How do you use Super Intelligence for meetings? Use SI before the meeting to assemble context and sharpen the agenda, during the meeting to capture state and evidence, and after the meeting to convert discussion into confirmed decisions, owners, deadlines and follow-through. The goal is not better transcripts. The goal is better coordination.
This article is part of the eduKateSG Workplace Super Intelligence Hub. It follows the Super Intelligence-Powered Workday and the dedicated Email workflow. This page owns the meeting layer: preparation, agenda design, notes, decisions, actions, handoffs and follow-through.
In this eduKateSG series, Super Intelligence is the practical machine-intelligence layer commonly described as artificial intelligence, generative AI, assistants, copilots, agents and connected automation. The strongest meeting use is not “AI takes notes for us.” It is “the meeting enters with clear state and leaves with clearer state.”
The Meeting Is a Workflow, Not an Event
A meeting has a trigger, purpose, participants, context, decisions, outputs and closure. Treating it as a calendar event alone hides the work around it. Preparation happens before. Follow-through happens after. Many meeting failures occur because these surrounding states are weak.
A useful meeting workflow is: Need → Prepare → Align → Decide → Assign → Record → Follow Through → Close. Super Intelligence can support every transition, but human responsibility for important decisions should remain visible.
The Four Jobs of a Meeting
- Information transfer: participants need shared facts or context.
- Decision: a choice must be made.
- Coordination: several owners need alignment around dependencies and next actions.
- Relationship: trust, negotiation, coaching, conflict or social context matters.
A meeting can contain more than one job, but it should be clear which job dominates. SI is especially strong at preparing information and preserving coordination state. Human presence remains particularly important where trust, judgment or authority is the core reason for the meeting.
The First Meeting Question: Should This Be a Meeting?
Before using SI to improve a meeting, ask whether the meeting should exist. If the objective is simply to share information, an asynchronous brief may be better. If the decision requires discussion, negotiation or collective judgment, a meeting may be appropriate.
Super Intelligence should not make meetings cheaper in ways that encourage more of them. Productivity sometimes means replacing the meeting with a clearer document, dashboard or decision request.
The Meeting Necessity Test
- Does a decision require real-time discussion?
- Does ambiguity need interactive clarification?
- Do several owners need to resolve dependencies together?
- Does the issue involve trust, negotiation or conflict?
- Would asynchronous communication create more delay than the meeting?
- Is there a defined output that justifies the participants’ time?
If most answers are no, redesign the work before optimising the meeting.
The Pre-Meeting Context Pack
A strong meeting starts before participants enter the room. SI can assemble a concise context pack from approved sources so the first half of the meeting is not spent reconstructing history.
- Objective: what must be true when the meeting ends?
- Current state: where does the work stand now?
- Previous decisions: what has already been agreed?
- Open questions: what remains unresolved?
- Evidence: which facts, documents or metrics matter?
- Constraints: what cannot change?
- Options: what credible alternatives exist?
- Decision owners: who has authority?
- Dependencies: who or what is blocked?
- Requested preparation: what should participants review in advance?
The pack should be short enough to use. Detailed evidence can be linked rather than copied into the main brief.
Meeting Preparation Should Reduce Reconstruction
The value of meeting preparation is not page count. It is the amount of reconstruction removed from live discussion. Participants should arrive knowing the shared state and where disagreement begins.
SI can compare the last meeting notes with current project state, identify commitments that remain open and surface changes since the previous decision. This is more useful than another general summary.
Agenda Design: Topics Are Too Weak
An agenda item called “Marketing Update” says little about why participants are present. A stronger agenda identifies the expected output: “Decide whether to delay launch by one week based on campaign readiness.”
Super Intelligence can turn topic-based agendas into output-based agendas by asking what decision, alignment or action each item requires.
The Decision-First Agenda
- Decision required
- Evidence required
- Decision owner
- Time box
- Expected options
- Known disagreement
- Next state after decision
Not every meeting needs a formal template, but consequential meetings benefit from decision-first structure.
The Meeting Brief
A practical SI-generated meeting brief can fit on one screen: purpose, top facts, previous commitments, decisions required, unresolved risks and requested preparation.
For senior participants, the brief should lead with consequence and decision rather than background. For technical participants, it may include more operational evidence. One state can support different audience views.
Participant Preparation
SI can produce role-specific preparation: what this participant needs to know, what questions they are likely to receive, which evidence they own and what decision authority they hold.
This reduces the common problem of participants entering with different assumptions about why they were invited.
The Pre-Meeting Question Set
- What outcome must this meeting produce?
- What would make the meeting unnecessary?
- Which facts are disputed?
- What decision is actually blocked?
- Who has legitimate authority to decide?
- What information must be current at the start?
- What should participants read beforehand?
- What should be handled asynchronously instead?
Participant Preparation Lab: Arrive With a Decision-Ready Brief
You are joining a meeting to contribute evidence and make a recommendation. Work through the complete packet below, then prepare the changed case before reading its answer. The useful result is a brief you can defend: each fact has a source, each question unlocks something, and each proposed commitment stays within the right person’s authority.
All people, organisations, dates, source records, prices and outputs in this lab are fictional. The SI drafts are constructed teaching examples, not results from a tested product. In this series, SI means practical AI assistance; this exercise makes no claim that present tools are artificial superintelligence. The labelled records are the complete approved evidence for the exercise. No real workplace material, recording, transcription, upload, invitation, order or message is needed. These fictional source rules do not grant permission to use or share a real meeting’s information.
Choose your starting point: read the source packet · inspect the completed preparation · repair the wrong brief · try the changed case · check your answer.
1. Source Packet: The Demonstration Folders
A1 · Approved meeting charter · Hana, operations lead · 5 October 2026, 16:00 SGT. The meeting is on 6 October, 10:00–10:20 SGT. Its purpose is to choose a feasible plan for 80 demonstration folders, each containing the approved version-3 insert, ready internally by 9 October at 12:00. Hana chairs and may approve this job’s printing expenditure up to S$100. Luis, procurement coordinator, may place an order only after Hana approves the expenditure and Arun approves the print proof. Mira, the participant you are preparing, owns the stock evidence and may recommend a plan. She cannot approve spending, release an order or change the promised contents or ready time. Hana, Luis, Arun and Mira will attend.
A2 · Confirmed specification · Hana · 4 October, 14:00. All 80 folders must contain a physical version-3 insert. Version 2 contains superseded information and cannot be substituted. A digital link does not fulfil this specification. The requirement remains in force unless Hana explicitly changes it. No change has been approved.
A3 · Old SI pre-brief · generated 4 October, 16:00. “There are 80 inserts in stock, so printing is unnecessary.” This summary came from an undifferentiated count. It was neither an approved specification nor a spending decision. It has not been refreshed.
A4 · Authoritative stock check for this job · Mira · 6 October, 09:10. Physical count: 80 inserts in total, comprising 52 undamaged version-3 inserts allocated to this job and 28 version-2 inserts. There is no other usable version-3 stock. This signed version-level count supersedes A3’s stock interpretation, while A2 still governs which version is usable.
A5 · Supplier quote verified by Luis · 6 October, 09:00. The supplier can print the missing 28 version-3 inserts. Standard service is S$2 per insert, delivered 9 October at 14:00. Rush service is S$3 per insert, delivered 8 October at 15:00 if the approved proof and order both reach the supplier by 6 October at 11:00. Prices include all charges for this exercise. The supplier has confirmed those quantities and arrival times under the stated conditions. No order has been placed.
A6 · Proof status · Arun, design owner · 6 October, 09:20. One correction to the version-3 print proof is still open. Arun has accepted the task of returning an approval or rejection by 10:40 today. A review deadline is not a guarantee that the proof will pass. No proof approval exists yet.
A7 · Assembly check · Mira · 6 October, 09:25. All other folder components are ready. A packing slot is reserved on 8 October, 16:00–17:00, with capacity for all 80 folders. Inserts must arrive by 16:00 for that slot. No additional slot or alternative stock has been confirmed. Assembly after a qualifying arrival would finish before the 9 October, 12:00 ready deadline.
A8 · Exercise handling rule · approved with the charter. A1–A7 may be used for this internal preparation exercise. The task is to draft Mira’s own brief, suggested agenda and questions. It does not include recording the meeting, distributing a brief, modifying a business system or purchasing anything. Keep any suggested new assignment labelled as a request until its proposed owner accepts it.
2. Work the Evidence Before Asking SI to Write
Resolve the apparent stock conflict. A3 and A4 both contain the number 80, but they answer different questions. A3 counts physical items; A4 identifies 52 that meet A2’s current specification. The usable-stock calculation is 80 required − 52 eligible = 28 inserts to print. The resolution rests on version-level evidence and the governing specification, not simply on picking the latest document.
Compare the actual options. Standard printing costs 28 × S$2 = S$56, but its 9 October, 14:00 arrival is two hours after the ready deadline and also misses the reserved packing slot. Rush printing costs 28 × S$3 = S$84, leaving S$16 within Hana’s S$100 limit. Rush delivery at 15:00 is one hour before the packing slot starts. It is a feasible candidate only if proof approval and order placement both happen before the 11:00 cutoff. Keeping the 28 version-2 inserts would fail A2 even though it avoids spending.
Keep the open dependency open. The desired meeting output is Hana’s decision on a conditional S$84 rush plan and an agreed response if proof approval is late or refused. It is not a claim that all folders are already ready. With the supplied evidence, no compliant fallback is confirmed. Mira may propose a fallback review; she may not invent spare stock, a later supplier cutoff or another person’s accepted action.
One bounded instruction for SI. “Using only A1–A8, prepare my brief as Mira, a four-item agenda totalling 20 minutes, and questions tied to unresolved decisions. Cite the record labels. Distinguish confirmed facts, my recommendation, existing accepted tasks and requests needing agreement. Calculate usable stock and costs. Preserve conditional deadlines. Do not write minutes for a meeting that has not happened or take any external action.” You can use the completed result below to check the response rather than judge it by how polished it sounds.
3. Completed Participant Brief, Agenda and Questions
Mira’s participant brief · prepared 6 October, 09:30 SGT. I recommend asking Hana to approve a conditional S$84 rush order for 28 version-3 inserts. We need 80 compliant folders ready by 9 October, 12:00. The current stock check confirms 52 usable inserts; the other 28 are obsolete version 2. The old “80 in stock” brief therefore overstates usable stock (A2–A4). Standard printing is S$56 but arrives too late. The S$84 rush option fits Hana’s S$100 authority and, if its conditions are met, arrives before our reserved packing slot (A1, A5, A7).
The outstanding condition is Arun’s proof approval. He has accepted a review deadline of 10:40; he has not approved the proof. Luis can order only after both proof approval and Hana’s expenditure approval, and the supplier needs both proof and order by 11:00 (A1, A5, A6). I will bring the signed version-level stock count and explain the 28-insert shortfall. I can recommend the plan but cannot authorise expenditure or order placement. I want the meeting to establish whether Hana approves the conditional plan, whether Luis accepts responsibility for acting within the cutoff once the conditions are met, and who will request a new decision if the proof fails or is late. No order, new assignment or successful delivery should be recorded as completed in advance.
Suggested agenda, subject to Hana’s agreement. 10:00–10:03: Mira reconciles 80 physical inserts with 52 usable inserts using A2–A4; output: shared shortfall of 28. 10:03–10:10: Hana compares standard, rush and no-print options using A5–A7; output: approve, reject or defer the conditional S$84 plan, with the condition stated. 10:10–10:16: Arun and Luis explain proof and order dependencies; output: confirm what each can accept and agree an escalation route if 10:40 passes without approval. 10:16–10:20: Hana reads back the actual decision and accepted responsibilities; output: a clear boundary between approval, pending proof, proposed actions and an order not yet placed. The timeboxes total 3 + 7 + 6 + 4 = 20 minutes.
Question for Hana: “Will you approve spending S$84 on the rush option, conditional on Arun approving the proof and Luis placing the order before 11:00?” This asks the authorised owner for the missing decision; it does not tell her that she has already agreed (A1, A5, A6).
Question for Arun: “Is the 10:40 approval-or-rejection response still achievable, and what exact correction remains open?” His answer may change feasibility. Asking whether the task is on track is different from assuming the proof will pass (A6).
Question for Luis: “If both approvals arrive in time, can you accept the order-placement step and confirm supplier receipt before 11:00?” A1 permits the action under conditions; it does not prove Luis has accepted this specific timed assignment or placed the order.
Question for Hana and the proposed escalation owner: “If the proof is rejected or unapproved at 10:40, who will bring you the blocked state immediately, and what may happen next without changing the specification or promised time silently?” This requests an accepted contingency process. The packet provides no confirmed compliant fallback.
My readiness check. I can show which source establishes the usable quantity, explain why the cheaper option fails, name both approvals, identify the supplier cutoff, and state the limits of my role without reading generated wording. If a new proof status or quote arrives before the meeting, I must update the affected conclusion and timestamp. Refreshing one field does not silently change the rest of the packet.
4. Repair a Polished Brief That Invents Commitments
Wrong draft, deliberately supplied for diagnosis: “All 80 inserts are ready. To be safe, Mira has approved a S$56 print top-up, and Luis will place it at 11:00. Arun will finish and approve the proof at 10:40. Standard delivery on 9 October meets the deadline. Everyone has agreed to the plan, so we can shorten the meeting to an update and confirm the completed order afterwards.”
Diagnosis. “All 80 ready” confuses physical stock with the version-3 requirement; A4 supports only 52 usable inserts. S$56 buys standard printing, whose 14:00 arrival fails the 12:00 deadline. Mira has no spending authority. Arun’s accepted review task may return rejection; it is not guaranteed approval. The supplier needs receipt by 11:00, so a promise merely to start ordering at 11:00 does not establish compliance. No source records Luis’s acceptance, group agreement or a completed order. The sentence about shortening the meeting removes the very approval and dependency questions the meeting exists to settle.
Complete repaired short brief: “We have 52 usable version-3 inserts for 80 required folders, leaving 28 to print. Standard printing costs S$56 but misses the ready deadline. I recommend asking Hana to approve the S$84 rush option conditionally. Arun must approve the proof, and Luis must accept and complete order placement with supplier receipt before 11:00. Arun’s promised 10:40 review may still return rejection. We need Hana’s actual decision and an accepted escalation route if approval is missing or refused. I can present the stock evidence and recommendation; I cannot approve expenditure or an order. No order or group agreement is confirmed.”
Check the repair against the source, not against the wrong draft. The repaired quantity follows A2/A4; the price and timing follow A5; the approval roles follow A1/A6. The remaining uncertainty is visible. A response that merely replaces S$56 with S$84 while keeping “everyone agreed” still fails. A response that removes every date to sound cautious also fails because it hides the decision’s time constraint.
5. Independent Preparation: The Authority and Stock Have Changed
Try this before reading the model answer. This is a separate fictional case, not an update to A1–A8. Use only B1–B7. Write a short participant brief, a three-item 15-minute agenda and four useful questions. Calculate the present shortfall, the smallest stock release that could make a rush order fit the chair’s authority, and the result if the release is refused. Label decisions and assignments that are still pending. Do not copy the S$84 recommendation from the first case.
B1 · Approved charter · 13 October 2026, 16:00 SGT. The 14 October meeting runs 10:00–10:15 SGT. The job is 60 demonstration folders with physical version-4 inserts, ready internally by 15 October, 12:00. Mira provides stock evidence and recommends. Dev chairs and may approve this job’s printing expenditure up to S$60. Hana alone may approve higher expenditure; she is unavailable until 12:00 on 14 October and has given no delegation. Luis may order after authorised spending approval and approved proof. Farah, who controls stock reserved for a separate job, will attend. Mira, Dev, Farah and Luis are the participants. No meeting participant can silently change this job’s contents or ready deadline.
B2 · West stock check · Mira · 14 October, 08:45. West has 36 usable version-4 inserts released to this job. East has another 18 usable version-4 inserts reserved for Farah’s separate job. None of those 18 is released to this job. The stock system’s “54 version-4 inserts across both sites” is a correct physical total, not an allocation of 54 to Mira’s job.
B3 · Farah’s stock note · 14 October, 09:00. Farah may release any whole number from 0 to 18 of the East inserts after reviewing the other job’s need. No release is yet approved. For this exercise, any approved release recorded by 10:30 will arrive at West by 15:00 using an already-approved transfer, at no additional charge. A release after 10:30 has no confirmed same-day arrival.
B4 · Supplier quote verified by Luis · 14 October, 09:05. Standard printing is S$2 per insert and arrives 15 October, 14:00. Rush printing is S$3 per insert and arrives 14 October, 16:00 if the approved proof and order are received by 11:00. Any positive whole-number quantity up to 24 is available at those unit prices; all charges are included. No order is placed, and no extension of the 11:00 cutoff is agreed.
B5 · Proof and assembly records · 14 October, 08:30 and 09:10. Arun has already approved the version-4 print proof. All other components are ready. The reserved packing slot is 14 October, 16:30–17:30, with capacity for all 60 folders; inserts arriving by 16:30 qualify. There is no confirmed later packing slot. No actual assembly is complete yet.
B6 · Generated pre-brief · 14 October, 09:15. “There are 54 usable inserts, so order six more for S$18. Dev’s attendance means the budget is approved.” This is an unreviewed summary of B1/B2, not evidence of allocation or spending approval.
B7 · Scope and permissions. Prepare Mira’s internal contribution using this fictional packet only. There is no authority to record, share, purchase, change a stock allocation or commit anyone to a new task as part of completing the exercise. A suggested agenda can ask for decisions; it cannot manufacture their answers.
6. Model Answer and Checking Rationale
Calculation. Only 36 inserts are currently allocated, so the present shortfall is 60 − 36 = 24. Rush-printing all 24 costs 24 × S$3 = S$72, which exceeds Dev’s S$60 authority by S$12. Dev can approve at most S$60 ÷ S$3 = 20 printed inserts. Therefore Farah must release at least 24 − 20 = 4 inserts before the rush route fits that limit. With exactly four released, the complete plan is 36 West + 4 transferred + 20 printed = 60, at S$60. This is a conditional option, not a prediction of Farah’s decision.
Completed participant brief · Mira · 14 October, 09:30 SGT. We currently have 36 allocated version-4 inserts for 60 required folders. The 18 at East remain reserved for Farah’s job, so B6’s six-insert top-up is unsupported. Printing the current 24-insert shortfall by rush service would cost S$72, outside Dev’s S$60 limit. I recommend first asking whether Farah can release at least four inserts by 10:30. If exactly four are released, Dev can consider a S$60 rush order for the remaining 20; a larger release would reduce the print quantity and cost. The approved transfer would arrive by 15:00 and rush printing by 16:00, both before the 16:30 packing slot (B2–B5).
The proof is already approved, but neither spending nor East-stock release is approved. Luis would still need to accept the timed order step and achieve supplier receipt by 11:00. I can provide the count and calculations; Farah decides the release and Dev decides expenditure within his limit. If fewer than four inserts are released, no complete rush plan fits Dev’s authority. Hana’s noon availability is after the supplier cutoff, and no extension is agreed. In that branch, the meeting should identify the blocked job and request an authorised alternative or change to the requirement, without promising that the existing deadline can be met. No order, East-stock release or packing completion has occurred (B1–B5).
Completed suggested agenda. 10:00–10:04: Mira and Farah distinguish physical stock from allocated stock and determine whether at least four East inserts can be released by 10:30; output: an actual release decision or a clearly pending/refused request. 10:04–10:10: Dev and Luis evaluate the resulting quantity, price and cutoff; output: an authorised conditional spending decision if within S$60, or a blocked state if not. 10:10–10:15: agree who will confirm any accepted release, supplier receipt and unresolved escalation; output: accepted next steps with conditions, without inventing an order. The timeboxes total 4 + 6 + 5 = 15 minutes.
Four questions. To Farah: “Can your other job spare at least four inserts, and can you authorise the release by 10:30?” To Dev: “Once the release quantity is confirmed, will you approve the remaining rush quantity at S$3 each, up to your S$60 limit?” To Luis: “If allocation and spending are approved, can you accept responsibility for getting the approved proof and order received before 11:00?” To Dev: “If fewer than four can be released, who will seek an authorised alternative or requirement change, given that Hana is unavailable before the present quote expires?” Each question changes feasibility or assigns an unresolved decision; none asks for facts already settled, such as whether the proof is approved.
Check the boundary cases. Zero released leaves 24 to print at S$72; three released leaves 21 at S$63; neither fits. Four released leaves 20 at S$60 and fits. All 18 released leaves six at S$18, but that is valid only after Farah’s release, which B6 skipped. Standard printing for the original 24 would cost S$48, but arrives at 14:00 on 15 October, two hours after the required ready time. Cheapness does not repair the deadline. A physical total can be accurate while its use as available stock is wrong.
If Farah refuses. Write: “No East stock is released. The current shortfall is 24 and rush cost is S$72, above Dev’s limit. With Hana unavailable until after the cutoff, the packet contains no approved, feasible complete plan. We need an authorised alternative or requirement change; the original deadline remains at risk.” It would be wrong to report the 60-folder job as approved, silently trim it to 56 folders, assume Hana will answer early or treat a request for a quote extension as a granted extension.
7. What Counts as Successful Preparation?
Compare your independent answer with five checks: it uses 36 allocated inserts rather than 54 physical inserts; calculates four as the minimum timely release; keeps stock release, spending approval, proof approval and order receipt separate; makes the 10:30 and 11:00 cutoffs visible; and explains the refused-release branch without inventing permission, stock or a changed deadline. Correct a failed check using the specific source record, then explain the repaired conclusion in your own words.
Finishing this fictional preparation task shows that you can produce and check a source-grounded participant contribution for these cases. It does not establish that an AI product saves meeting time, that a real team approved the proposal or that the workplace process is safe to automate. Keep using the article’s existing meeting workflow and its recording, privacy and permission boundaries when moving from this preparation lab to a real situation.
During the Meeting: Capture State, Not Every Word
A transcript can be useful, but the durable meeting record should be structured around state. What was decided? What changed? What remains unresolved? Who owns the next action? What evidence matters?
A transcript is raw material. A decision log and action register are operating artefacts.
The Five Live Capture Layers
- Facts: material information established during discussion.
- Questions: unresolved issues that affect progress.
- Decisions: choices actually confirmed.
- Actions: next steps with owner and deadline.
- Risks: conditions that could change the plan.
SI can extract candidates in real time or immediately afterwards, but material decisions should be confirmed by the chair or authorised owner.
Decision vs Discussion
One of the most common meeting failures is treating a discussed idea as a decision. The record should distinguish: proposed, discussed, recommended, approved, rejected and deferred.
Super Intelligence can identify decision language, but the final state should be confirmed by the people with authority.
The Decision Record
- Decision
- Owner
- Date
- Evidence considered
- Alternatives rejected
- Important assumptions
- Constraints
- Trigger for reconsideration
- Required follow-up
This record makes future meetings shorter because the organisation no longer has to debate what was previously decided.
The Action Record
- Action
- Owner
- Deadline or trigger
- Dependency
- Required evidence
- Destination system
- Closure condition
An action without an owner is a wish. An action without a deadline or trigger is easy to forget. An action without closure can remain permanently “in progress.”
The Parking Lot
Meetings need a place for issues that matter but do not belong to the current decision. SI can capture these in a parking-lot list with owner and next forum.
The parking lot prevents scope drift without discarding useful questions.
Useful Live Assistance During Meetings
For decision meetings, SI can challenge the current proposal before or during discussion by surfacing counterarguments, alternative explanations, failure scenarios and evidence that would change the recommendation. This is useful when it exposes weak assumptions early rather than manufacturing opposition for its own sake.
Before a high-value meeting, confirm that the people present actually hold the required decision authority, expertise and evidence. If the person who must approve cannot attend, the meeting may only be able to prepare a recommendation. Super Intelligence cannot replace missing institutional authority.
During discussion, use SI selectively. It can retrieve a fact or document, but consequential claims should still be checked against the authoritative source. Use deterministic calculators or spreadsheets for exact arithmetic, with SI explaining implications around the verified numbers rather than substituting persuasive prose for calculation.
SI can also generate scenarios or options quickly when the group needs alternatives, and it can assist with translation, terminology clarification or accessible summaries where approved. Generated scenarios remain proposals for evaluation, while translation and accessibility support should match the meeting’s confidentiality and precision requirements.
The Meeting Timebox
SI can help propose timeboxes based on the expected output of each agenda item. Information items should usually receive less time than decisions or negotiation.
The chair should still adapt when important new evidence appears. Timeboxing is a coordination tool, not a reason to suppress necessary discussion.
The Meeting Chair and Super Intelligence
SI can prepare prompts, track agenda progress, capture unresolved items and draft the decision log. The chair remains responsible for participation, conflict, decision legitimacy and whether the room has enough evidence to close the issue.
The Participant and Super Intelligence
Participants can use SI before the meeting to prepare questions, review evidence and clarify terminology. During the meeting, attention should remain on the human discussion rather than on continuous side conversations with the assistant.
The Note-Taker and Super Intelligence
SI can dramatically reduce manual note-taking, but the record should be reviewed for attribution, commitments and decisions. Automated notes should not make participants less attentive to what they are actually agreeing to.
The Observer and Super Intelligence
For complex workshops, SI can help track themes, repeated objections and unresolved questions. This is useful for facilitation but should not be treated as an objective psychological reading of participants.
Post-Meeting Closure Should Happen Quickly
The longer the gap between discussion and structured follow-up, the more context is lost. A strong workflow prepares the decision log and action record immediately after the meeting while the state is fresh.
SI can create a first-pass closure packet within minutes. The chair or owner confirms material items before distribution.
The Post-Meeting Closure Packet
- Purpose and outcome
- Decisions confirmed
- Actions and owners
- Deadlines and triggers
- Unresolved questions
- Risks
- Dependencies
- Next checkpoint
- Links to evidence
- Changes to project or system state
The Meeting-to-Task Handoff
Confirmed actions should leave the meeting record and enter the actual task or project system. A transcript archive is not a task-management system.
SI can prepare tasks and owners, but the target system should return the actual created state.
The Meeting-to-Calendar Handoff
If a meeting creates another event or deadline, update the calendar rather than leaving the commitment in prose. The confirmed event becomes the source of truth.
The Meeting-to-Email Handoff
Some follow-up belongs in email. SI can draft concise messages from the decision record instead of requiring the sender to reconstruct the meeting manually.
The dedicated email article covers commitment and follow-up controls in more depth.
The Meeting-to-Knowledge Handoff
A meeting may create durable organisational knowledge: a new procedure, policy interpretation, product decision or lesson learned. Move that knowledge into the maintained source rather than leaving it buried in minutes.
The Meeting-to-Handoff Handoff
If responsibility moves between teams after the meeting, create a structured handoff packet rather than forwarding the transcript. Use state, evidence, uncertainty, next action, owner and closure.
See How Super Intelligence Can Improve Handoffs.
The Follow-Through Monitor
SI can watch whether actions are accepted, completed, overdue or blocked. This is more useful than sending generic reminders.
A follow-through alert should explain what is overdue, why it matters, who owns it and what dependency is affected.
The Next-Meeting Preparation Loop
Before the next recurring meeting, SI can compare previous commitments with current state. This creates a closed loop: decision → action → monitoring → next review.
The meeting becomes an operating checkpoint rather than a repeated status-reconstruction session.
Meeting Type: One-on-One
SI can help prepare previous commitments, goals, feedback examples and open questions. The human conversation remains central because coaching, trust and sensitive context matter.
The post-meeting record should distinguish confidential notes from commitments that belong in shared systems.
Meeting Type: Project Status
A project meeting should focus on exceptions, decisions and dependencies rather than reading status aloud. SI can prepare the current state from project tools and highlight changes.
This can shorten the meeting and increase decision density.
Meeting Type: Executive Review
Executive meetings benefit from concise evidence, decision framing and risk. SI can prepare an exception brief and record confirmed decisions while executives retain strategic authority.
Meeting Type: Client Meeting
SI can prepare account history, open commitments and agenda questions. After the meeting, it can draft follow-up from confirmed commitments.
Commercial promises, sensitive tone and relationship strategy remain human-led.
Meeting Type: Sales Call
Before the call, SI can create account and opportunity context. During or after, it can identify needs, objections, commitments and next steps. The salesperson should confirm any pricing or commercial commitment.
Meeting Type: Incident Meeting
Incident meetings need current state, confirmed facts, hypotheses, owners and next checks. SI can assemble timelines and action logs, but operational authority and public communication remain governed.
Meeting Type: Hiring Interview Debrief
SI can organise interviewer notes against explicit job-related criteria, but hiring decisions remain fairness-sensitive and should not be delegated blindly.
Meeting Type: Legal or Contract Review
SI can compare documents, track issues and record approved positions. Qualified professionals retain legal interpretation and commitment authority.
Meeting Type: Education or Student Review
SI can organise learning evidence, prior interventions and open questions. Educators retain pedagogical, welfare and high-stakes judgment.
Meeting Type: Creative Workshop
SI can generate alternatives, cluster ideas and capture themes. The human group decides which ideas are valuable and how they fit purpose, taste and constraints.
Meeting Type: Retrospective
SI can summarise patterns, incidents, blockers and repeated causes. The team should validate causal claims and decide which process changes deserve action.
Meeting Type: Board or Governance Meeting
SI can prepare evidence and draft minutes, but formal decisions, fiduciary responsibilities and governance requirements remain with the authorised body and applicable process.
The Meeting Metrics
- Preparation time
- Time spent reconstructing context
- Decision count
- Decision latency
- Action acceptance rate
- Action completion rate
- Clarification after meeting
- Repeat discussion rate
- Percentage of agenda items with explicit outcomes
- Meeting duration
- Follow-up latency
- Participant time saved
Do not use attendance or transcript length as primary productivity metrics.
Decision Density
Decision density asks how much of the meeting contributes to meaningful decisions or coordination. SI can improve density by moving information transfer into the pre-read and bringing evidence into the room.
Repeat Discussion Rate
If the same issue appears in multiple meetings because previous decisions or actions were not captured, meeting memory is weak. A structured decision log can reduce this rate.
Action Completion Rate
Actions are the bridge from meeting to work. Track whether confirmed actions are actually accepted and completed.
Clarification Rate
High post-meeting clarification indicates that decisions, ownership or next actions were not clear enough.
Meeting Load
Measure total participant hours, not just meeting count. A one-hour meeting with ten people consumes ten person-hours before preparation and recovery.
Meeting Necessity Rate
Review recurring meetings periodically. Some should be shortened, made asynchronous or cancelled once shared state improves.
The Meeting Automation Ladder
- Manual meeting brief
- SI-assisted agenda
- Automated context assembly
- Automated note capture
- SI decision/action extraction
- Human-confirmed closure packet
- Automatic task creation after approval
- Condition-based action monitoring
- Automatic preparation for the next recurring meeting
Each level should be justified by measurable reduction in coordination friction.
The Meeting Privacy Boundary
Recording, transcription and meeting analysis may involve personal or confidential information. Use approved tools and follow applicable organisational policies, consent requirements and confidentiality obligations.
Not every meeting should be recorded merely because the technology makes it easy.
The Sensitive Meeting Boundary
Personnel, legal, medical, security and other sensitive meetings may require stronger restrictions on recording, storage, access and automated analysis.
SI can still support preparation or private drafting where appropriate without creating a permanent transcript.
The Meeting Hallucination Boundary
Automated minutes can misattribute statements or infer decisions that were never made. Material records should be reviewed against the actual discussion or confirmed by responsible participants.
The Automation-Bias Boundary
Participants may trust polished meeting summaries and stop checking them. Use decision confirmation and source links for consequential items.
The Meeting Data-Minimisation Rule
Capture what is needed for coordination and accountability. Do not preserve every conversational detail by default, especially when the information is sensitive or temporary.
The Meeting Source-of-Truth Rule
The meeting record can document the decision, but project, CRM, contract, finance or other business systems should remain authoritative for operational state.
The Meeting Currentness Rule
Pre-read information should be fresh enough for the decision. A meeting based on stale project or customer state can make a well-reasoned decision wrong.
The Meeting Authority Rule
SI should not infer who can decide merely from who spoke most. Authority comes from organisational role, governance and the meeting’s explicit decision design.
The Meeting Follow-Up Rule
A meeting is not complete when minutes are sent. It is complete when decisions and actions enter the real workflow with owners and closure conditions.
The Meeting Simplification Rule
If SI makes information available asynchronously, remove the status-reading portion of the meeting. Use live time for ambiguity, trade-offs, relationships and decisions.
The Meeting Cancellation Rule
If the expected decision can be made asynchronously with sufficient evidence and no meaningful discussion is required, cancel or replace the meeting.
The Meeting Human-Agency Rule
Participants should remain able to correct the record, reject an inferred decision and clarify uncertainty. SI should make the meeting more legible, not freeze a mistaken interpretation into the workflow.
The Meeting Learning Loop
Recurring meetings create repeated evidence about what works. If the same agenda item always overruns, the same question remains unresolved or the same follow-up fails, redesign the meeting system.
SI can surface these patterns across meeting records, but the team decides what to change.
A 30-Day Meeting Improvement Build
Week 1 — Observe
Collect four or five meetings and record preparation, purpose, decisions, actions and follow-up.
Week 2 — Prepare
Introduce SI-generated context packs and decision-first agendas.
Week 3 — Close
Introduce structured decision and action records with human confirmation.
Week 4 — Monitor
Track follow-up and review whether meetings can be shortened or replaced.
The Meeting Review Checklist
- Meeting has a clear purpose.
- Expected decisions or outputs are explicit.
- Participants are necessary.
- Context is available before the meeting.
- Current data is fresh.
- Discussion and decision are distinguished.
- Actions have owners and deadlines.
- Unresolved questions remain visible.
- Sensitive information is handled appropriately.
- Follow-up enters authoritative systems.
- Next checkpoint is known.
- Recurring meeting necessity is periodically reviewed.
What This Article Owns
This page owns meetings as a workplace Super Intelligence workflow: preparation, agendas, note capture, decisions, actions, follow-through, privacy boundaries and measurement.
The broader workday article owns daily operating rhythm; the handoff article owns cross-boundary state transfer; the email article owns asynchronous message queues.
Frequently Asked Questions
Can Super Intelligence take meeting notes?
Yes. Notes are most useful when they are transformed into confirmed decisions, actions, owners, deadlines, unresolved questions and evidence rather than stored as raw transcript alone.
Can SI prepare meeting agendas?
Yes. Strong agendas are decision- or output-oriented and use current context from approved sources.
Should every meeting be recorded?
No. Recording and transcription should follow organisational policy, confidentiality, consent and the actual need for a durable record.
Can SI decide what the meeting agreed?
It can identify candidate decisions, but material decisions should be confirmed by the authorised participants.
Can SI create tasks automatically after meetings?
Yes, after the actions and owners are confirmed and the destination system is appropriate. Task creation is an external action and should return observable status.
Can SI shorten meetings?
Often. Move information reconstruction into pre-meeting briefs and use live time for decisions, exceptions and relationships.
Can SI replace meetings?
Some information-sharing meetings can become asynchronous. Meetings involving negotiation, ambiguity, trust or collective judgment may still be valuable.
How do I measure better meetings?
Measure preparation time, decision latency, repeat discussion, clarification, action completion, follow-up latency and total participant time.
What comes next?
Continue to How to Use Super Intelligence for Professional Writing, which moves from meeting state into the production of clear, evidence-grounded workplace documents.
The Core Meeting Rule
Use Super Intelligence to make meetings start with shared context and end with shared state. Everything in between should justify the human time in the room.
The strongest meeting system does not create better transcripts. It creates fewer repeated discussions, clearer decisions, better handoffs and more reliable follow-through.
Edge Case: Meeting With No Decision Owner
Sometimes a group meets to discuss an issue but nobody present can approve the next step. SI can identify this before the meeting by comparing the agenda with role authority. The repair is to include the right owner or explicitly redefine the meeting as recommendation preparation.
This prevents the common pattern where a group appears aligned but must schedule another meeting for the actual decision.
Edge Case: Meeting With Conflicting Sources
Two teams may arrive with different figures, policy versions or project states. SI can surface the conflict, but the meeting should not silently choose one. Identify the authoritative source or assign reconciliation before the decision.
Edge Case: Meeting With Missing Evidence
If an important assumption cannot be verified, the meeting can decide under uncertainty only if participants understand the gap. Otherwise, defer the decision and assign the missing evidence.
A polished SI brief should never make missing evidence look present.
Edge Case: Meeting With External Participants
External customers, suppliers or partners change the privacy and communication boundary. Confirm what may be recorded, what documents may be shared and what commitments the organisation is authorised to make.
Role-specific post-meeting summaries may be safer than distributing the full internal record.
Edge Case: Meeting With Sensitive Personal Content
Performance, welfare, health and disciplinary conversations may require stricter access and retention. SI can support preparation without automatically creating a detailed permanent transcript.
Edge Case: Meeting During an Incident
Time pressure can produce rapid decisions based on partial evidence. SI should distinguish confirmed state, hypothesis and action. The incident leader remains responsible for operational authority.
Edge Case: Meeting Across Time Zones
Some participants may be disadvantaged by recurring scheduling. Better asynchronous briefs and decision packets can reduce the number of times people must attend live outside normal hours.
Edge Case: Meeting With Language Differences
Translation and terminology support can improve inclusion, but technical or contractual meaning should be verified where precision matters.
Edge Case: Meeting With Too Many Actions
When SI extracts dozens of tasks, stop and prioritise. A meeting should not create more work than the organisation can absorb. Confirm the few actions that materially advance the outcome.
Edge Case: Meeting With No Follow-Up System
If the organisation has no project or task tool, use a simple shared action register. The essential requirement is visible ownership and closure, not sophisticated software.
Edge Case: Meeting With Repeated Reopening
If the same decision returns, inspect whether rationale, assumptions or ownership were missing. Reopening may also be legitimate when a reassessment trigger occurred.
The Meeting Deletion Framework
A meeting should be considered for deletion when its primary function can be replaced by clearer asynchronous state. Deletion does not mean communication stops; it means the same outcome is achieved with less synchronous cost.
- Status is already visible in a dashboard or project system.
- No decision requires simultaneous discussion.
- Participants rarely contribute.
- Actions can be assigned asynchronously.
- The meeting routinely ends early or without decisions.
- The same information is repeated elsewhere.
- The meeting exists mainly because no one prepared a brief.
The Meeting Shortening Framework
If the meeting still has synchronous value, shorten it by moving context and status out. Begin with the decision or exception immediately.
SI-generated pre-reads make shorter meetings practical because participants can arrive with shared context.
The Meeting Participant-Reduction Framework
Keep decision-makers, evidence owners and execution owners. Move observers to the post-meeting brief unless their live presence adds relationship or judgment value.
The Meeting Frequency-Reduction Framework
If weekly issues are sparse, move to fortnightly or exception-triggered cadence. Use SI monitoring to surface when the conditions for a live meeting actually appear.
The Meeting Exception-Only Framework
Routine updates remain asynchronous. Live time occurs only when a threshold, blocker, conflict or major decision requires it.
This is a mature use of SI because intelligence monitors the normal state while humans gather for the abnormal state.
The Meeting Removal Test After 30 Days
For one recurring meeting, run a four-week experiment with asynchronous state briefs and live exception review. Measure total participant hours, missed decisions, action closure and stakeholder satisfaction.
If coordination remains strong with fewer live hours, preserve the new design.
The Meeting Recovery Plan
If automated notes, tasks or follow-up are wrong, the team should know how to correct the shared state. A human-confirmed decision log should remain the recovery anchor.
The Meeting Outage Plan
If SI is unavailable, participants should still be able to read the agenda, record decisions and actions manually and update project systems. The process can become slower without becoming undefined.
The Meeting Incident Review
For a material meeting-system failure, preserve the transcript or source material where appropriate, the generated summary, confirmed corrections, affected actions and external communications. Then identify whether the root cause was capture, interpretation, review or tool action.
The Meeting Correction Workflow
Participants should have a clear path to correct a misrecorded decision or action. Corrections should update the canonical meeting record and downstream task state rather than living in another email thread.
The Meeting Dispute Workflow
If participants disagree about what was decided, return to the confirmed decision owner or evidence rather than allowing the model summary to arbitrate the disagreement.
The Meeting Transfer-to-Team Workflow
When meeting output affects a wider team, prepare a role-appropriate brief: what changed, what remains unchanged, what action is required and where questions should go.
This reduces the tendency to invite everyone to the meeting for awareness.
The Meeting Transfer-to-Leadership Workflow
Leadership may need only decision, risk, unresolved issue and resource request. SI can compress operational detail without erasing uncertainty.
The Meeting Transfer-to-Customer Workflow
External follow-up should include confirmed commitments and next steps, not internal speculation. Human review remains important for promises, pricing and relationship-sensitive language.
The Meeting Transfer-to-Documentation Workflow
When a recurring issue is resolved, update the SOP, FAQ, architecture decision record or policy where appropriate. SI can draft the change, but the responsible owner verifies it.
The Meeting Knowledge-Debt Signal
If meetings repeatedly explain the same concept, the organisation has knowledge debt. Capture the explanation once and make it accessible.
This is one of the ways SI can help reduce meeting demand over time.
The Meeting Coordination-Debt Signal
If meetings repeatedly exist to discover who owns what, the organisation has ownership or handoff debt. Fix the operating interface rather than adding better minutes.
The Meeting Decision-Debt Signal
If meetings repeatedly postpone decisions because evidence is missing or authority is absent, improve decision preparation and invite the correct owner.
The Meeting Action-Debt Signal
If actions repeatedly remain incomplete, strengthen task acceptance, deadlines and monitoring rather than scheduling another meeting to ask for status.
The Meeting Data-Debt Signal
If teams meet to reconcile conflicting numbers, repair the source-of-truth and data process. SI can help diagnose mismatches but should not institutionalise repeated reconciliation.
The Meeting Risk-Debt Signal
If important meetings rely on informal memory of previous decisions, build a decision log. Organisational memory should not depend on who happened to attend.
The Meeting Accessibility Benefit
Well-structured meeting briefs, summaries and transcripts can improve access for people who process information differently or cannot attend. The design should provide flexible views without forcing everyone into live participation.
The Meeting Search Benefit
SI can make past meeting decisions easier to retrieve by project, topic, owner or date. Search should return the canonical decision record and source, not only an unverified generated summary.
The Meeting Continuity Benefit
When team members join or return after absence, structured meeting state reduces the cost of catching up. This is especially valuable in long projects and distributed teams.
The Meeting Learning Benefit
Retrospective analysis across meetings can reveal recurring blockers, unresolved decisions and repeated knowledge gaps. Use these patterns to improve the organisation rather than merely summarise history.
The Meeting Cost-Benefit Review
Once per quarter, choose the most expensive recurring meetings by participant hours. Compare their decision yield, action closure and relationship value. The goal is to identify where better preparation, fewer participants or async replacement will create the largest return.
The Meeting Portfolio
Treat recurring meetings as a portfolio. Each should have purpose, owner, cadence, participant logic and expected output. Obsolete meetings should be retired like obsolete software.
The Meeting Charter
- Purpose
- Expected outputs
- Decision owner
- Core participants
- Optional participants
- Pre-read standard
- Agenda structure
- Action destination
- Cadence
- Review or sunset date
A charter is especially useful for expensive recurring meetings.
The Meeting Owner
Every recurring meeting should have an owner responsible for its continued usefulness. Ownership includes cancelling when the meeting no longer serves its purpose.
The Meeting Sunset Date
Set a date to review whether a recurring meeting should continue. This prevents calendar inertia.
The Meeting Audit Trail
For high-impact decisions, preserve enough evidence to reconstruct the decision, owner and rationale. For low-risk routine meetings, lighter documentation is sufficient.
The Meeting Data-Minimisation Rule
Retain only what the workflow needs. A full transcript can contain far more sensitive information than the final decision record requires.
The Meeting Least-Privilege Rule
If an assistant can create tasks, send email or update systems, scope those permissions to the meeting workflow. Listening to a meeting should not automatically grant broad action rights.
The Meeting Human-Confirmation Rule
Material decisions, commitments and sensitive tasks should be confirmed by the authorised person before the system treats them as final state.
The Meeting Error-Correction Rule
Make corrections visible and propagate them to downstream systems. A corrected summary is insufficient if the wrong task or message already remains active elsewhere.
The Meeting Learning-Loop Rule
Repeated problems should change the meeting system: better pre-read, different participants, shorter agenda, clearer decision rights or meeting removal.
The Meeting Final Checklist
- Do we need synchronous interaction?
- What outcome must exist at the end?
- Who can decide?
- What evidence must be read first?
- Who truly needs to attend?
- What can move asynchronous?
- How will decisions be confirmed?
- How will actions be accepted?
- Where will the state live after the meeting?
- What follow-up should be monitored?
- When will we review whether this meeting should still exist?
The Final Meeting Principle
Super Intelligence should make meetings shorter, fewer, better prepared and harder to forget.
If SI only creates transcripts of the same low-value meetings, the workplace has automated memory without improving coordination. The real gain comes when synchronous time is reserved for decisions, relationships and interaction that genuinely need people together.
The Meeting Decision Review
Important decisions deserve a later review when outcomes become visible. Compare what the group believed at the time, which assumptions were recorded, what happened and whether the decision process missed information.
SI can connect later outcomes back to the original decision log, creating a learning loop across meetings instead of treating every meeting as a separate event.
The Meeting Assumption Monitor
Some decisions remain valid only while assumptions remain true. A launch depends on supplier delivery; a budget depends on demand; a hiring plan depends on approved headcount.
An SI workflow can monitor those assumptions and surface the meeting decision for reconsideration when one changes.
The Meeting Commitment Monitor
Actions and promises should move into a monitored system. SI can watch for overdue commitments and bring back only the items requiring attention.
This reduces the common pattern where the next meeting is spent rediscovering actions that were already agreed.
The Meeting Dependency Monitor
Cross-team dependencies often determine whether meeting decisions can be executed. Track which action is blocked by another team, system or approval and who owns the unblock.
Monitoring dependencies can be more valuable than generating more detailed meeting minutes.
The Meeting Knowledge Loop
Recurring explanations, definitions and decisions should improve the knowledge base. If every new employee asks the same question in meetings, capture the accepted answer and source it properly.
Meetings become less repetitive when organisational learning leaves the room.
The Meeting Feedback Loop
Ask participants whether the new SI-enabled meeting process improves preparation, live focus and follow-up. Combine that with measured outcomes such as attendance hours, action completion and clarification.
Participant enthusiasm can reveal usability, while workflow metrics reveal whether the system actually works.
The Meeting Attention Budget
Every meeting consumes shared attention. Ten participants in a one-hour meeting consume ten person-hours before preparation and recovery. This simple arithmetic helps organisations see why deleting low-value meetings can create more capacity than improving note-taking.
SI should be used to spend shared attention more deliberately, not to make meeting cost less visible.
The Meeting Recovery Cost
After a meeting, participants often need time to recover context before returning to deep work. Dense meeting schedules create fragmentation even when each meeting is short.
Meeting productivity should therefore consider calendar structure and context switching, not only duration.
The Meeting Cluster Strategy
Where practical, cluster meetings into defined windows and preserve larger focus blocks elsewhere in the day. SI preparation can reduce the time required between meetings to reconstruct context.
The right pattern depends on role; executives and managers may require more synchronous work than makers or analysts.
The Meeting Prep Automation Boundary
Pre-meeting preparation is generally safer to automate than live decision-making because the output remains reviewable before the meeting. This makes it an excellent starting point.
Automate source gathering and brief structure first, then decide whether role-specific views or scheduling support add enough value.
The Meeting Note Automation Boundary
Automatic note capture can be useful when policy permits, but the system should not transform tentative discussion into confirmed state without review.
The closer a note is to a consequential decision or commitment, the stronger the confirmation requirement.
The Meeting Action Automation Boundary
Creating tasks automatically is appropriate only after action extraction is reliable and owners can accept or reject. The destination system should return the created task state.
The Meeting Follow-Up Automation Boundary
Routine follow-up messages can be prepared automatically, but external commitments, sensitive issues and unresolved facts should remain human-reviewed.
The Meeting Scheduling Automation Boundary
SI can propose scheduling based on availability and urgency, but recurring meetings should not be created automatically without a clear continuing trigger.
Calendar convenience can otherwise produce meeting inflation.
The Meeting Deletion Automation Boundary
A system can flag meetings with low decision rate, low participation or repetitive status content as candidates for review. The human owner decides whether to delete, shorten or convert them asynchronously.
The Meeting Accessibility Layer
SI can help generate concise summaries, translations or accessible notes for participants who need them. The organisation should preserve accuracy and confidentiality appropriate to the content.
The Meeting Cross-Time-Zone Layer
Distributed teams benefit from better asynchronous pre-reads and post-meeting packets. A person unable to attend should be able to understand confirmed decisions and act without replaying the entire meeting.
This can reduce the number of duplicate meetings held for different time zones.
The Meeting Hybrid-Work Layer
Hybrid meetings can disadvantage remote participants when side conversations or physical-room cues are lost. Structured agendas and explicit decision confirmation can reduce this information gap.
SI notes may help, but facilitation remains a human responsibility.
The Meeting Confidential-Session Layer
Some sessions should use no recording or automatic notes at all. Sensitive personnel, legal, security or strategic discussions may require tighter handling.
A mature SI meeting system includes the ability to turn automation off easily and visibly.
The Meeting Legal-Discovery Layer
Recording and transcript retention can have legal or compliance implications depending on the organisation and jurisdiction. Meeting technology should follow the relevant policy and legal guidance rather than treating storage as free.
The Meeting Personal-Data Layer
Meeting notes can contain personal data and sensitive assessments. Limit access and capture to what the meeting purpose requires.
For Singapore organisations, applicable PDPA obligations and PDPC guidance should be considered where personal data is processed by AI-enabled systems.
The Meeting Agentic Layer
A meeting agent might retrieve context, propose agenda, capture decisions, create tasks and monitor follow-up. That is several distinct tool permissions combined into one workflow.
Grant them progressively. An agent that can observe and prepare can prove value before it receives authority to write to systems or send messages.
The Meeting Multi-Agent Layer
Complex organisations might use separate agents for transcription, knowledge retrieval, task routing or compliance checks. Multiple agents can add coordination overhead and conflicting state.
Use them only when role separation creates real value and shared state can be governed.
The Meeting Failure Drill
Test the system with a meeting where participants discuss an action but never approve it, where an owner changes mid-meeting, where evidence conflicts, where sensitive information appears and where a tool update fails.
The output should preserve ambiguity, require confirmation where needed and avoid claiming completion without world-return evidence.
The Meeting Regression Set
Keep a few representative meetings or synthetic examples covering routine decisions, ambiguous actions, dissent, sensitive content and no-decision outcomes. Re-run them after major system changes.
The Meeting Change Ledger
For important automated meeting workflows, record material changes to transcription, summarisation prompts, model, source integrations and action routing. This helps explain why behaviour changes over time.
The Meeting Portfolio Audit
At team or department level, review recurring meetings by purpose, participant hours, decision output and follow-through. Identify which can be removed, combined, shortened or made asynchronous.
This is where SI can create organisational leverage beyond any single meeting.
The Meeting Portfolio Categories
- Keep: live interaction clearly creates value.
- Shorten: preparation can move async.
- Shrink: fewer participants are needed.
- Reduce frequency: trigger does not justify current cadence.
- Convert async: primarily information transfer.
- Delete: outcome would not materially worsen.
The Meeting Change-Management Rule
When meeting workflows change, tell participants which old work disappears. If SI produces the pre-read and action log, stop asking people to create duplicate notes or status decks.
Productivity requires retiring redundant work, not layering automation on top.
The Meeting Cultural Rule
Participants should be allowed to say “not decided”, “unknown” and “needs more evidence”. An intelligent summary should not pressure the group toward artificial closure.
Good meeting culture values clear uncertainty over polished false consensus.
The Meeting Relationship Rule
Some meetings create value precisely through human presence: trust, empathy, coaching and conflict resolution. SI should reduce background work around them while leaving space for the interaction itself.
The Meeting Creativity Rule
Use machine-generated ideas as stimulus, not as the boundary of imagination. Human participants should have time to generate independent ideas before seeing AI suggestions in creative sessions.
The Meeting Professional-Judgment Rule
In legal, medical, financial, educational, safety or other specialist meetings, SI can prepare evidence and records, but professional judgment remains under the appropriate qualified processes.
The Meeting Executive-Authority Rule
Executives and boards can use SI for decision support, but formal authority and accountability remain with the people and institutions designated to exercise them.
The Meeting Final Checklist
- Do we need a meeting?
- What job does the meeting perform?
- What output must exist at the end?
- Who has authority to decide?
- What context should arrive before the meeting?
- Which evidence should be visible?
- What can be handled asynchronously?
- How will discussion be distinguished from decision?
- How will actions be confirmed?
- Where will accepted state be stored?
- Who monitors follow-through?
- When is the next meeting actually required?
Frequently Asked Questions: Advanced Meeting Use
Can SI attend a meeting for me?
It can capture or summarise where policy and tools permit, but it cannot always replace your relationship, authority or responsibility. Use an AI proxy only when the meeting’s purpose is compatible with asynchronous representation.
Can SI vote in a meeting?
It can provide decision support or analysis, but institutional decision rights belong to the authorised people or governed systems. Treat model output as evidence or recommendation unless a specific process legitimately grants more authority.
Can SI detect decisions automatically?
It can identify candidate decision statements, but material decisions should be confirmed because conversational language is often tentative or ambiguous.
Can SI detect action items automatically?
Yes, but action ownership and deadlines should be confirmed before the system creates binding commitments.
Can SI cancel unnecessary meetings?
It can flag candidates based on purpose, agenda, participant role and historical output, but the owner should decide whether relationship or context still justifies the session.
Can SI create meeting minutes automatically?
Yes. Structured minutes focusing on decisions, actions and unresolved questions are generally more operationally useful than exhaustive prose.
Can SI monitor meeting actions?
Yes. Condition-based monitoring can surface overdue commitments and changed dependencies without requiring constant manual checking.
Can SI prepare different meeting summaries for different teams?
Yes, provided all versions remain grounded in the same confirmed state. Role-specific views should not create contradictory facts.
How much meeting content should be stored?
Store what is justified by purpose, policy, privacy, retention and operational need. More storage is not automatically more useful.
How do we know when to stop improving a meeting?
When the meeting reliably achieves its purpose with acceptable preparation, participant cost and follow-through, further automation may create little value. Maturity includes leaving a good process alone.
The Meeting Rule in One Sentence
Move information out of the room, keep judgment and interaction in the room, and move confirmed state back into the operating system immediately afterward.
That is the practical role of Super Intelligence in meetings: less reconstruction, better preparation, clearer decisions and follow-through that survives beyond the conversation.
Meeting Archetype 1 — The Decision Meeting
A decision meeting should begin with an explicit decision statement: what choice must be made, by whom, using which evidence, within what constraints and by when. SI can prepare the option matrix, summarise evidence, identify assumptions and surface unanswered questions.
The live meeting should not spend most of its time reading the packet. It should challenge assumptions, resolve disagreement and make the decision. After the meeting, SI can draft the decision record while the authorised decision owner confirms it.
Meeting Archetype 2 — The Coordination Meeting
Coordination meetings exist because several people or teams must align timing, dependencies and ownership. SI can assemble current project state, identify blocked dependencies and create a proposed action map before the session.
The live meeting then concentrates on dependency resolution rather than status recitation. The closure packet should update owners, dates and blockers in the real project system.
Meeting Archetype 3 — The Problem-Solving Meeting
Problem-solving meetings benefit from a clearly framed problem, known evidence, constraints and unresolved hypotheses. SI can prepare a chronology, cluster symptoms, propose hypotheses and identify missing data.
The group should distinguish facts from hypotheses. A polished machine explanation should not be promoted to root cause without evidence.
Meeting Archetype 4 — The Brainstorm
For brainstorming, SI can widen the option space before or during the session. It can generate alternatives, combine themes and challenge obvious ideas. The human group should still protect originality by not anchoring solely on machine-generated suggestions.
One useful pattern is human-first ideation followed by SI expansion, then human evaluation. Another is SI stimulus before the meeting followed by human divergence. The best sequence depends on whether machine suggestions would help or bias the group.
Meeting Archetype 5 — The Review
A review meeting inspects work against criteria. SI can compare the artefact with the rubric, highlight changes and assemble evidence. Reviewers then focus on judgment, trade-offs and exceptions.
This pattern works in project reviews, document reviews, code reviews, academic review, finance and quality assurance.
Meeting Archetype 6 — The Retrospective
A retrospective asks what happened, why and what should change. SI can assemble timelines, cluster recurring themes and compare with previous retrospectives. Human participants should validate causal claims and decide which behaviours or system changes deserve action.
The value of the retrospective is not the summary; it is whether repeated failure creates a different future process.
Meeting Archetype 7 — The One-on-One
SI can prepare the previous commitments, current goals, open questions and relevant evidence. This can help the manager or employee arrive with continuity.
The conversation itself remains relational. Coaching, trust, motivation, sensitive feedback and personal nuance should not be reduced to generated scripts or scores.
Meeting Archetype 8 — The Client Meeting
Before the meeting, SI can prepare account history, previous commitments, current project state and suggested questions. After the meeting, it can draft follow-up from confirmed decisions and commitments.
The human retains commercial judgment, trust and any promise that affects scope, price, delivery or relationship.
Meeting Archetype 9 — The Executive Review
Executives need decision-relevant compression. SI can prepare risks, scenarios, dependencies and unanswered questions from a large evidence base.
The meeting should focus on decisions, trade-offs and direction rather than on repeating operational detail. The closure should create an explicit decision and delegation state.
Meeting Archetype 10 — The Cross-Functional Handoff
When one team transfers responsibility to another, the meeting is an interface. SI can prepare the handoff packet: current state, evidence, commitments, dependencies, unresolved issues and next owner.
A successful meeting ends when the receiving team accepts ownership and the shared system records the transfer.
The Meeting Input Standard
A meeting should not begin with scattered context. Important inputs should be available before the session: current project state, relevant documents, prior decisions, metrics, customer or stakeholder context and specific unresolved questions.
SI can assemble the inputs, but the meeting owner should decide which materials are authoritative and current.
The Meeting Output Standard
Every meeting should have an explicit output standard. Depending on the meeting, that may be a decision, a prioritised plan, a resolved dependency, a commitment, a set of accepted actions or a documented unresolved question.
If the meeting ends with only a summary, the output may be weaker than the purpose required.
The Meeting Acceptance Standard
The meeting owner should know what “done” means. A decision meeting is done when the authorised decision is recorded. A coordination meeting is done when owners and dependencies are clear. A one-on-one may be done when the intended conversation occurred and commitments are understood.
This avoids the vague outcome of “we talked about it.”
Meeting Failure Mode: No Shared State
Participants arrive with different assumptions about current status. The first half of the meeting becomes reconstruction.
Repair with a pre-meeting state brief generated from approved sources and reviewed for currentness.
Meeting Failure Mode: No Decision Owner
The group discusses an issue extensively but nobody in the room can make the final decision. The meeting creates a recommendation, not closure.
Repair by inviting the appropriate authority or redefining the output honestly as preparation for a later decision.
Meeting Failure Mode: Too Many Participants
Participants attend because information might be relevant, not because their role is required. This multiplies time cost and can reduce candour.
Use SI-generated asynchronous briefs for informational stakeholders and keep the live meeting for the people whose interaction is necessary.
Meeting Failure Mode: Status Recitation
Everyone reads out updates that could have been written. The meeting consumes live time without using the advantage of synchronous interaction.
Repair by generating the status asynchronously and reserving live time for blockers, disagreement and decision.
Meeting Failure Mode: Transcript Dependence
The organisation records everything but nobody converts the transcript into operating state. Search becomes easier, but decisions and actions remain hidden.
Repair by making the decision log, action log and unresolved-question list the primary outputs.
Meeting Failure Mode: Inferred Commitments
SI interprets conversational language as an action or promise and assigns it to someone who did not accept it.
Repair by requiring owner confirmation for material commitments before they enter project systems.
Meeting Failure Mode: No Follow-Through
The action list exists but nobody monitors it. The same issues reappear next week.
Repair with condition-based follow-up, clear owners and closure signals.
Meeting Failure Mode: Decision Drift
A decision is made but later participants remember it differently. The lack of an authoritative decision record forces the organisation to reopen the issue.
Repair by confirming material decisions quickly and storing them in the appropriate shared record.
Meeting Failure Mode: Stale Pre-Reads
The meeting makes a decision based on information that changed after the pre-read was generated.
Repair by refreshing time-sensitive facts before the session and marking the timestamp of material data.
Meeting Failure Mode: AI Overreach
The system not only summarises discussion but also infers motives, consensus or psychological states. These interpretations can be unreliable and inappropriate.
Keep SI focused on observable content, decisions, actions and evidence rather than speculative readings of people.
Meeting Failure Mode: Recording Everything
Continuous recording creates privacy, storage and behavioural costs. Some conversations work better without a permanent transcript.
Capture only what the meeting’s operating purpose and applicable policies justify.
Meeting Failure Mode: Agenda Explosion
SI identifies every open issue and places them all on the agenda. The meeting becomes impossible to complete.
Prioritise agenda items by decision need, consequence and dependency. Low-priority items can be handled asynchronously or deferred.
Meeting Failure Mode: Meeting Creation as Default
The system sees a conflict or open question and automatically schedules a meeting. This can create calendar inflation.
Use a decision rule: meeting only when synchronous interaction is more useful than asynchronous resolution.
Meeting Failure Mode: No Receiver
The closure packet is generated but no project owner, task system or downstream team receives it. The meeting produces documentation without movement.
Every material action or decision should have a destination.
The Meeting Review Capacity Problem
If every meeting produces a long AI-generated summary that participants must review, meeting automation can create another queue. Keep the review object short and focused on material state.
The transcript may remain available for reference without requiring line-by-line approval.
The Meeting Information Architecture
Separate meeting artefacts by purpose: brief, agenda, evidence, transcript, decision log, action log and closure packet. A single giant document tries to serve too many functions.
SI can transform between these artefacts while preserving source links.
The Meeting Source Hierarchy
- Business system: authoritative current state.
- Approved document: policy, contract or formal reference.
- Decision log: confirmed prior choices.
- Transcript: record of discussion.
- AI summary: navigation layer over the evidence.
The hierarchy helps participants know what to trust when sources conflict.
The Meeting Currentness Rule
For time-sensitive decisions, the meeting should know how fresh important facts must be. Inventory, account state, pricing, project status or incident severity may need same-day or real-time refresh.
A correct summary of stale facts can still produce the wrong decision.
The Meeting Confidentiality Rule
Not every participant should receive every detail. Role-specific briefs can minimise sensitive exposure while preserving the shared state needed for the decision.
Access controls should be enforced by systems rather than by prompt instructions alone.
The Meeting Retention Rule
Decide how long recordings, transcripts and generated notes should be kept according to organisational policy, legal requirements and the actual need for the data.
More storage is not automatically more organisational memory.
The Meeting Consent Rule
Where required by law, policy or professional norms, participants should know when recording or transcription is occurring. Meeting technology should not surprise the people whose conversation is being captured.
The Meeting Personal-Data Rule
Personnel, customer, student, patient or other personal information deserves appropriate access and handling. Minimise what is captured and who can retrieve it.
The Meeting Security Rule
If meeting content can trigger tool actions, treat external or participant-supplied text as untrusted content. A statement in a meeting should not override the system’s real permission boundary.
The Meeting Tool-Use Rule
Generating a task, sending email or editing project state are separate permissions from note capture. Grant these progressively and require confirmation where consequence justifies it.
The Meeting World-Return Rule
When SI creates a task, event or record after a meeting, require confirmation from the destination system. The action list is not the same as a created task.
The Meeting Version Rule
If meeting briefs or decision logs change after distribution, preserve version or update history where the decision is important. Participants should know which record is current.
The Meeting Agent Pattern
A bounded meeting agent can assemble pre-read, propose agenda, capture structured notes, prepare actions and monitor follow-through. It should not silently decide what participants agreed or schedule new commitments without appropriate confirmation.
The agent role is coordination support, not meeting authority.
The Recurring Meeting Agent
For a recurring operating review, an agent can compare last meeting commitments with current state and build an exception brief automatically.
This is high leverage because it turns the recurring meeting into a control loop rather than a repeated reconstruction exercise.
The Project Review Agent
A project-review agent can retrieve milestone state, identify deviations, summarise risks and prepare decisions required. The project owner verifies state before the meeting.
The One-on-One Assistant
A one-on-one assistant can remind the manager of previous commitments, goals and open topics. It should not infer employee sentiment or performance conclusions from sparse data.
The Client Meeting Assistant
It can prepare account history, open actions and recent communications, then draft follow-up afterwards. Sensitive commercial decisions remain with the responsible person.
The Research Meeting Assistant
It can prepare source summaries, competing findings and unresolved methodological questions. Researchers retain judgment over evidence quality and interpretation.
The Learning Meeting Assistant
For training or education, SI can prepare learner evidence, questions and next-step options. Teachers or trainers remain responsible for learning decisions and sensitive context.
The Meeting Cost Model
Meeting cost includes participant time, preparation, context switching and follow-up. A 45-minute meeting with eight participants is at least six person-hours before any preparation or downstream work.
SI can reduce preparation and follow-up, but eliminating an unnecessary meeting often produces the largest return.
The Meeting Time-to-Value Model
Meeting SI often has fast time-to-value because preparation and closure can improve before deep integration. A simple context brief and structured decision record can create value immediately.
More advanced tool actions and recurring agents can follow after the meeting state model is stable.
The Meeting Pilot
Choose one recurring meeting with visible friction. Baseline preparation time, meeting duration, repeat discussion, action ambiguity and follow-up latency. Introduce SI first for pre-read and closure.
Keep decision authority unchanged. Measure whether the live conversation becomes shorter, more decisive or better prepared.
The Meeting Pilot Case Set
- Normal meeting with complete context
- Meeting with missing pre-read
- Meeting where decision authority is absent
- Meeting with conflicting evidence
- Meeting with several action owners
- Meeting with sensitive content
- Meeting that should have been asynchronous
The last case is especially important: a successful SI meeting system should sometimes recommend no meeting.
The Meeting Experiment: Pre-Read Only
Start by generating one-page pre-reads and compare preparation time, repeated context questions and decision latency. This is low risk and can prove whether context reconstruction is the problem.
The Meeting Experiment: Closure Only
Keep the meeting unchanged but use SI to create decision and action records afterward. Measure follow-up clarity, action acceptance and repeat discussion next time.
The Meeting Experiment: End-to-End
Only after the components work should the organisation connect pre-read, agenda, note capture, closure and monitoring into one meeting workflow.
The Meeting Promotion Gate
Move toward automatic task creation or recurring-agent support only when decisions and actions are already extracted reliably and humans confirm the material state efficiently.
The Meeting Demotion Gate
Reduce automation when misattribution, missed nuance, privacy concerns or reviewer overload increases. Returning to manual confirmation is a valid operating response.
The Meeting Retirement Gate
Retire a meeting entirely when the work can be handled asynchronously without harming decision quality, coordination or relationships.
The Meeting Portfolio
Managers can review recurring meetings across a team: purpose, cadence, participant hours, decision output and follow-through. SI can surface meetings with low decision density or repeated unresolved topics.
The portfolio view helps organisations improve the calendar as a system rather than one meeting at a time.
The Meeting Hygiene Checklist
- Purpose is explicit.
- Meeting necessity has been tested.
- Expected outcome is defined.
- Participants are required for that outcome.
- Context arrives beforehand.
- Agenda is decision- or action-oriented.
- Current evidence is available.
- Decisions are confirmed explicitly.
- Actions have owners and deadlines.
- Sensitive data is handled appropriately.
- Closure enters real systems.
- Recurring necessity is periodically reassessed.
The Meeting Owner Checklist
- Confirm objective.
- Confirm decision authority.
- Approve pre-read.
- Start with current state.
- Protect the agenda.
- Confirm decisions.
- Confirm actions.
- Approve closure packet.
- Ensure system updates.
- Review follow-through.
The Participant Checklist
- Read the brief.
- Bring evidence you own.
- State uncertainty clearly.
- Challenge assumptions.
- Do not imply commitment you cannot make.
- Confirm actions assigned to you.
- Correct the record if needed.
The SI Checklist
- Use approved sources.
- Preserve timestamps and versions.
- Separate fact from inference.
- Do not infer authority.
- Flag uncertainty.
- Do not convert discussion into decision automatically.
- Return external action status.
- Escalate sensitive or unsupported cases.
The Final Meeting Operating Standard
A mature SI-enabled meeting can answer five questions immediately after it ends: What was decided? What remains unresolved? Who owns the next actions? What evidence matters? When will the next state be checked?
If participants still need to reread the transcript to answer those questions, the meeting system is not yet operating at full value.
The Final Meeting Principle
The purpose of Super Intelligence in meetings is to compress context before people meet and preserve state after they separate.
Live human time should be reserved for what benefits from being live and human: challenge, judgment, trust, negotiation, creativity and legitimate decision-making.
The Meeting Operating System
A mature meeting workflow connects preparation, synchronous interaction and follow-through into one operating loop. The meeting itself is only the middle stage. The system should know what triggered the meeting, what output is required, which evidence matters, who has authority, what state changed and where accepted actions live afterward.
This operating-system view prevents meeting technology from being reduced to transcription. Transcription captures words. A meeting operating system captures state.
The Six Meeting States
- Proposed: a meeting may be needed but purpose is not yet confirmed.
- Prepared: objective, agenda, participants and evidence are ready.
- In session: synchronous discussion is occurring.
- Decision pending: discussion occurred but authority or evidence is still incomplete.
- Closed: confirmed decisions and actions are recorded.
- Monitored: follow-through and reassessment conditions are being tracked.
These states make it easier to know what Super Intelligence should do next and when it should stop.
Meeting Trigger Design
A meeting should have a trigger stronger than calendar habit. Triggers include a decision deadline, threshold breach, cross-team dependency, unresolved disagreement, customer event or need for live relationship work.
Recurring meetings should periodically prove that their trigger still exists. If not, delete or convert them.
Meeting Entry Criteria
A meeting is ready to start when the purpose is clear, relevant decision owners are available, necessary context exists and the expected output is understood. Starting without those conditions often produces another meeting rather than a decision.
SI can run an entry check and flag missing preconditions before participants spend shared time.
Meeting Exit Criteria
A meeting is complete when the required decision, alignment, creative output or relationship outcome has been achieved and any actions are assigned. If the purpose remains unresolved, the record should say why and what evidence or authority is still missing.
Ending at the scheduled time is not the same as achieving the exit criteria.
The Meeting Preparation Depth Test
Preparation should match consequence. A routine team sync may need a short state brief. A board decision, legal negotiation or incident review may require a full decision pack with evidence, assumptions and risks.
SI can generate both, but the meeting owner should choose the depth instead of allowing every session to accumulate excessive pre-reading.
The Meeting Evidence Map
For each agenda item, identify the source or evidence that matters: project state, financial figures, customer feedback, contract language, logs, research or prior decisions.
This makes live fact-checking faster and prevents the meeting from depending on memory.
The Meeting Decision-Authority Map
Record who can recommend, who can decide, who can veto where relevant and who executes. The same person may hold several roles, but they should not remain implicit.
If the decision owner is absent, the meeting should state whether it is preparing a recommendation rather than pretending closure.
The Meeting Participation Map
Classify participants by role: decision owner, evidence owner, execution owner, relationship participant, facilitator or observer. Observers can often receive the post-meeting brief instead of attending.
This is one of the simplest ways to reduce meeting cost without reducing decision quality.
The Meeting Attention Map
Different agenda items demand different levels of cognitive attention. Put the most consequential decisions before fatigue, not after forty minutes of status updates.
SI can help reorder the agenda based on decision value, dependencies and readiness.
The Meeting Dependency Map
A decision may depend on another team, document, approval or technical check. Make those dependencies visible before discussion. Otherwise participants can spend time debating an option that cannot yet be executed.
The Meeting Assumption Register
Important assumptions should be captured explicitly. Examples include launch date, demand forecast, budget availability, customer need, technical feasibility or legal interpretation.
A later monitoring workflow can watch these assumptions and reopen the decision when they change.
The Meeting Risk Register
For high-impact meetings, maintain a concise list of risks raised, owner and mitigation. SI can extract candidate risks, but humans decide materiality.
The Meeting Dissent Register
When a consequential decision has material dissent, record it where appropriate. Future review becomes stronger when teams can see whether a minority concern later proved important.
This is more useful than a summary that creates artificial unanimity.
The Meeting Open-Evidence Register
If a claim needs verification before action, record the evidence request, owner and deadline. The meeting can close without pretending the fact has been established.
The Meeting Decision Freeze
Once a decision is confirmed, freeze the exact wording in the decision log. Later summaries can explain it, but the confirmed statement remains stable.
This prevents AI-generated paraphrases from gradually changing organisational meaning.
The Meeting Action Acceptance Step
Action items should be accepted by the owner where possible. The owner may adjust deadline or clarify scope before the action becomes active.
This small confirmation reduces the common problem of assignments created in the room but never truly owned.
The Meeting Action Definition of Done
Every consequential action should have a completion state. “Review contract” may be too vague. “Return approved redlines to Procurement” is clearer.
SI can help turn vague actions into testable outcomes.
The Meeting Action Dependency Check
Before assigning an action, check whether the owner has required information, access and upstream decisions. A task that begins blocked should say so explicitly.
The Meeting Action Destination
Accepted actions belong in a project, task, CRM, ticket or other operating system. Meeting notes can link to them but should not remain the only place they exist.
The Meeting Decision Destination
Material decisions should move to the place future teams will look: project decision log, board record, policy repository or other maintained system.
The meeting transcript is evidence; the confirmed record is state.
The Meeting Knowledge Destination
If a meeting establishes a reusable answer or procedure, create or update the canonical knowledge source. SI can draft the update from the meeting evidence.
The Meeting Relationship Destination
Not every outcome belongs in software. Trust, alignment and mutual understanding may be the main result. Avoid forcing sensitive interpersonal outcomes into reductive metrics simply because SI can summarise them.
Meeting Productivity for Individuals
For an individual, meeting SI should reduce preparation time, note-taking burden and post-meeting reconstruction. It should also make it easier to decline meetings whose purpose does not require attendance.
Meeting Productivity for Teams
For a team, the biggest value often comes from shared state: everyone sees the same confirmed decisions, owners and open questions.
This reduces duplicate note-taking and later arguments about what was agreed.
Meeting Productivity for Managers
Managers can use SI to prepare one-on-ones, surface blocked commitments and close action loops. They should retain responsibility for feedback, prioritisation and sensitive decisions.
Meeting Productivity for Executives
Executives benefit from decision compression: fewer raw updates, more clear risks and choices. SI should help them enter the meeting at the decision edge rather than at the start of the information chain.
Meeting Productivity for Distributed Teams
Distributed teams benefit from role-specific async briefs, decision logs and post-meeting packets. SI can reduce the pressure for duplicate meetings across time zones.
Meeting Productivity for Cross-Functional Work
Cross-functional meetings have the highest translation cost because terminology, incentives and systems differ. SI can translate the same canonical state into the language each function needs while preserving shared facts.
Meeting Preparation Failure: Too Much Context
A large generated brief can overwhelm participants. Repair by leading with the decision, changes and risks, and link deeper sources below.
Meeting Preparation Failure: Wrong Context
The system retrieves old or irrelevant material. Repair source selection and currentness rather than asking participants to distrust every brief.
Meeting Preparation Failure: Missing Decision Owner
The meeting is well prepared but cannot close. Repair attendance or redefine the output as recommendation.
Meeting Preparation Failure: Generic Agenda
The system generates standard topics unrelated to the real state. Repair by grounding agenda design in the workflow and required decision.
In-Meeting Failure: Transcript Dominance
Participants assume the transcript will capture everything and become less deliberate about decisions. Repair by confirming decision statements and actions live.
In-Meeting Failure: False Consensus
SI summary smooths disagreement. Repair by preserving material dissent and unresolved issues.
In-Meeting Failure: AI Interruptions
The assistant surfaces too many suggestions or facts, fragmenting the human conversation. Repair by using SI on request or at defined checkpoints.
In-Meeting Failure: Over-Reliance on Instant Answers
Participants accept generated facts under time pressure. Repair by showing sources and marking unverified claims.
Post-Meeting Failure: Notes Without Action
A polished summary is distributed but nothing enters the task system. Repair the closure workflow.
Post-Meeting Failure: Actions Without Acceptance
The system creates tasks that owners do not recognise. Repair with human confirmation or a clear team convention.
Post-Meeting Failure: Duplicate Systems
Meeting notes, task tools and project systems all hold different versions of state. Repair by choosing authoritative destinations and using notes as links or context.
Post-Meeting Failure: Over-Automated Follow-Up
The system sends messages or schedules events based on inferred commitments. Repair by requiring confirmation for consequential actions.
Meeting Governance: Low Risk
Routine internal meetings may need only approved transcription, participant awareness, human confirmation of actions and appropriate storage.
Meeting Governance: Medium Risk
Client, financial or cross-functional decision meetings need stronger source handling, access control, action confirmation and record ownership.
Meeting Governance: High Impact
Legal, employment, security, board, safety or other high-impact meetings may require restricted recording, specialist ownership, stronger retention rules and formal decision records.
Meeting Privacy in Practice
Meeting automation can capture people who never intended every statement to become permanent. Use purpose limitation, access control and sensible retention.
Where recording or transcription requires notice or consent, follow the applicable policy and law.
Meeting Security in Practice
Meeting content can include credentials, incident details, confidential strategy or customer data. Avoid exposing unnecessary content to AI systems and keep action permissions separate from note-taking.
Meeting Agent Security
If an agent can act on meeting outputs, treat participant messages and shared documents as potentially untrusted content. Trusted workflow rules should determine which actions are permitted.
Meeting Data Minimisation
Record the minimum content necessary for the meeting’s purpose and follow-up. Not every joke, aside or speculative thought belongs in permanent organisational memory.
Meeting Retention Design
Different artefacts can have different lifetimes. The action log may remain while the raw transcript expires sooner. The correct design depends on business, legal and privacy needs.
Meeting Access Design
A broad transcript may contain information inappropriate for all action owners. Create role-appropriate summaries while controlling access to deeper evidence.
Meeting Change Control
If the system automatically extracts decisions or actions, material changes to model, prompt or integration should be regression-tested against representative meeting examples.
Meeting Outage Fallback
The meeting should remain operable when SI is unavailable. A human can capture decisions and actions manually; the system may reduce effort but should not be the only way the organisation knows what was decided.
Meeting Review Capacity
If the organisation records hundreds of meetings, nobody can manually verify full summaries. Focus human confirmation on decisions, actions, commitments and sensitive state rather than every sentence.
Meeting Exception Handling
Allow participants or organisers to mark a meeting as restricted, no-recording, manual-review or no-automation. One default should not govern every meeting class.
Meeting Readiness Audit Questions
- Is the meeting purpose explicit?
- Can the required output be named?
- Are decision owners present?
- Are sources current?
- Is automated capture appropriate?
- Can participants correct the record?
- Are actions confirmed before routing?
- Are destination systems defined?
- Can follow-up be monitored?
- Can the meeting run safely without SI?
Meeting Pilot Design
Choose one recurring meeting class with clear pain. Measure preparation time, participant hours, decision rate, action clarity and follow-up delay. Start with pre-meeting briefs and post-meeting closure before adding live intervention.
This keeps the pilot low risk and makes value easier to attribute.
Meeting Pilot Success Criteria
- Preparation time falls.
- Participants spend less live time reconstructing history.
- Decision statements are clearer.
- Action ownership improves.
- Clarification after the meeting decreases.
- Follow-up enters operating systems faster.
- Low-value meetings can be shortened or removed.
Meeting Pilot Failure Criteria
- Review of AI notes costs more than manual capture.
- Important nuance is repeatedly lost.
- Participants trust incorrect summaries.
- Actions are created without acceptance.
- Meeting count rises because automation makes meetings feel cheaper.
- Privacy or retention burden outweighs value.
Meeting Improvement: Repair
Repair when the meeting itself is useful but preparation, notes or follow-up are weak. Strengthen the specific layer rather than redesigning everything.
Meeting Improvement: Stabilise
Standardise agenda, decision and action structures across one meeting class. Use examples and templates so several organisers can achieve similar quality.
Meeting Improvement: Automate
Once state is stable, automate source retrieval, brief generation, task preparation or monitoring. Keep authority boundaries explicit.
Meeting Improvement: Delete
Delete when the objective can be met asynchronously or no longer matters. The most productive automated meeting is sometimes the one the organisation no longer needs.
The Meeting Portfolio Review
Once a quarter, review recurring meeting classes: total participant hours, decision output, action completion, async potential and whether the trigger still exists.
SI can analyse patterns across calendars and notes, but managers should decide which meetings still create value.
The Meeting Standard for Teams
A team-level standard can be simple: every meeting has purpose, expected output, owner, brief where needed, confirmed decisions, confirmed actions and a destination for state.
This creates enough structure for SI assistance without forcing every meeting into bureaucracy.
The Meeting Standard for Individuals
An individual should know which meetings require preparation, which can be declined, what evidence to bring, what commitments were accepted and what follow-up to monitor.
SI can maintain continuity across those states inside the broader personal workday.
The Meeting Standard for Organisations
At scale, organisations should know which meeting technologies are approved, how recordings and transcripts are handled, how actions enter systems, and how sensitive meeting classes are treated.
This becomes part of workplace SI governance as meeting intelligence moves from personal notes to shared infrastructure.
The Final Meeting Decision
Before each recurring meeting, the owner should be able to choose one of five states: keep, shorten, shrink, convert async or delete.
Super Intelligence provides the evidence and coordination support; the organisation decides how much synchronous time is worth spending.
The Final Meeting Rule
Move information out of the room, keep human interaction in the room when it adds value, and move confirmed state back into the operating system immediately afterward.
That is how Super Intelligence makes meetings better: fewer minutes spent reconstructing, more time spent deciding and relating, and far less ambiguity after everyone leaves.
The Meeting Before-and-After Comparison
Compare the workflow before and after Super Intelligence is introduced. Before: participants search for context, status is reconstructed live, decisions are buried in discussion, actions are written in different formats and follow-up depends on memory. After: context is prepared beforehand, live time is used for interaction, confirmed decisions are explicit, actions enter real systems and unresolved items are monitored.
The comparison should identify which burden disappeared and which new control appeared. If the only change is a more polished transcript, the meeting has not been redesigned deeply enough.
The Meeting Receiver Test
Ask the people who depend on meeting output whether they receive better state. Can project teams act without clarification? Can leaders understand the decision without reopening the entire discussion? Can absent stakeholders see the current state without being invited next time?
Receiver effort is part of meeting productivity because meeting value continues after the room empties.
The Meeting Independence Test
Ask whether a participant who missed the meeting can reconstruct the decision, evidence, owner and next action from the accepted record. If not, too much meaning remains in private memory.
The objective is not to replace human presence in every sensitive meeting. It is to make operational state durable enough that work can continue.
The Meeting Resilience Test
Ask what happens if transcription fails, the SI service is unavailable or a participant disputes the generated record. The meeting should still be operable. Human confirmation, source evidence and fallback note-taking remain important for consequential meetings.
The Meeting Scale Test
A workflow that works for one team may fail across the company because meeting formats, privacy requirements, project systems and decision rights differ. Standardise the grammar—context, decision, action, owner, evidence, closure—while allowing local fields and controls to vary.
The Meeting Transfer Test
Before copying an SI meeting workflow to another function, compare the meeting job, data sensitivity, receiver, authority and consequences. A technical design review and a personnel meeting may both use structured notes, but they should not share the same recording, retention or automation assumptions.
The Meeting Audit Questions
- Does the meeting have a clear job?
- Could part or all of it be asynchronous?
- Do participants receive current context beforehand?
- Is decision authority present?
- Are facts, recommendations and decisions separated?
- Do actions have owners and closure?
- Does follow-up enter real systems?
- Are privacy and retention appropriate?
- Does the recurring meeting still deserve its place on the calendar?
The Meeting Governance Review
Recurring or high-impact meeting workflows should have clear rules for recording, access, retention, task creation and action authority. A meeting assistant that only drafts notes is materially different from an agent that can update project systems, send follow-up and schedule the next event. Governance should follow the actual behaviour of the system rather than the product label.
The meeting owner should know which steps remain human-confirmed, which fields come from authoritative systems and which actions can be reversed. When those boundaries change, the workflow should be reviewed again.
The Meeting Portfolio Review
Once several recurring meetings use SI, review the portfolio rather than treating each session independently. Which meetings have high decision density? Which repeatedly reopen the same issues? Which generate actions that are not completed? Which exist mainly to distribute information that could be asynchronous?
Super Intelligence can surface these patterns across calendars, agendas and closure records. Management can then simplify the meeting system itself: cancel low-value sessions, shorten status-heavy meetings, improve pre-reads and reserve live time for interaction that genuinely benefits from people being together.
The Meeting Learning Flywheel
Every repeated meeting failure should become a design signal. Missing context means the pre-read is weak. Repeated clarification means decisions or actions are ambiguous. Overdue actions mean ownership or follow-up is weak. Reopened decisions mean the decision record is insufficient or the original assumptions changed.
Capture these signals and improve the meeting workflow. The strongest recurring meetings become shorter and more decisive over time because the organisation learns how to prepare, decide and close more effectively.
The Meeting Closure Rule
A meeting is complete only when its important state survives outside the meeting. A confirmed decision should be findable. An action should have an owner. A deadline should exist in the correct system. An unresolved question should have a route. A follow-up should be monitorable.
Super Intelligence earns its place when it makes that closure faster and more reliable without pretending to own the human judgment that created the decision.
The Meeting Operating Templates
A mature SI meeting system uses a small set of reusable templates rather than inventing a new prompt for every meeting. The templates should reflect the work that must move before, during and after the conversation.
Template 1 — Decision brief
- Decision required
- Current state
- Evidence
- Options
- Trade-offs
- Unknowns
- Decision owner
- Deadline
The brief should fit on one screen or page where possible. Detailed evidence remains linked rather than copied into the brief.
Template 2 — Meeting agenda
- Objective
- Decision or output for each item
- Owner
- Evidence required
- Time box
- Close state
A close state might be decided, assigned for evidence, escalated, deferred with condition or closed with no action.
Template 3 — Decision record
- Decision
- Owner
- Date
- Evidence
- Rationale
- Assumptions
- Actions
- Reassessment trigger
The decision record becomes the durable organisational memory; the transcript remains supporting evidence when needed.
Template 4 — Action record
- Action
- Owner
- Due date
- Dependency
- Definition of done
- Destination system
- Status
Action records should move into the project or task system rather than remain inside meeting notes.
Template 5 — Open-question register
- Question
- Why it matters
- Evidence needed
- Owner
- Due date
- Decision blocked
Open questions should remain visible until resolved. SI should not convert them into guessed answers simply to make the meeting record complete.
The 60-Minute-to-25-Minute Meeting Redesign
A common workplace opportunity is not removing a meeting completely but compressing it. Begin by moving status and evidence into an asynchronous brief. Participants read or scan before the meeting. The live session starts with the highest-value unresolved issue.
- Collect routine status asynchronously.
- Use SI to normalise updates and surface exceptions.
- Circulate the decision brief before the meeting.
- Reduce attendees to decision, evidence and execution roles.
- Open with the decision or blocker.
- Time-box discussion.
- Confirm decision and actions live.
- Close early when the output is achieved.
The meeting may shrink from sixty minutes to twenty-five without losing coordination because the information work has moved outside the synchronous window.
The 25-Minute-to-Zero Meeting Redesign
Some meetings can disappear entirely after several cycles of good asynchronous state. If participants already agree on the status, no exception exists and the decision owner can approve from a written brief, there may be no reason to gather live.
Use an exception trigger: the meeting occurs only when a threshold, unresolved conflict or consequential decision appears. SI can monitor for that condition and prepare the context when the meeting becomes necessary.
Meeting Economics: Person-Hours, Not Calendar Blocks
Meeting cost should be calculated in participant hours rather than organiser calendar time. A weekly one-hour meeting with twelve participants consumes twelve person-hours each week. Over forty-eight working weeks, that is 576 person-hours before preparation, switching and follow-up.
If SI can move status updates asynchronous and reduce the meeting to thirty minutes with six essential participants, the annual synchronous cost falls dramatically. The real value is recovered attention, not merely shorter calendar entries.
Meeting Economics: Decision Delay
A slow meeting cadence can create a second cost: decisions wait for the next scheduled session. If a weekly committee is the only place a decision can happen, work may sit for days.
A decision brief plus asynchronous approval or condition-triggered meeting can reduce time-to-decision without increasing meeting volume.
Meeting Economics: Follow-Up Debt
Poor closure creates hidden cost after the meeting. Participants send clarification emails, reconstruct actions and schedule another meeting to discover what happened.
SI creates value when it reduces this follow-up debt through confirmed decision and action records.
Meeting Economics: Rework
If participants leave with different interpretations, teams may execute conflicting work. The cost can exceed the meeting itself. Preserve one confirmed current state and make corrections propagate to downstream systems.
The Meeting Quality Ladder
Level 0 — Calendar event
People gather, talk and leave. No structured output exists.
Level 1 — Notes
The meeting has a written record but decisions and actions may still be ambiguous.
Level 2 — Decision and action capture
The meeting produces explicit decisions, owners and deadlines.
Level 3 — Prepared meeting
Participants receive current context and evidence before live discussion.
Level 4 — Connected meeting
Confirmed state returns to project, CRM, task or other systems.
Level 5 — Exception-driven meeting system
Routine work is asynchronous; meetings occur only when live interaction creates real value.
The ladder is not a status contest. A one-off conversation may need only Level 1. Expensive recurring governance meetings may deserve Level 5 discipline.
The Meeting Portfolio Maturity Model
An organisation with many recurring meetings should manage them as a portfolio. Each meeting has purpose, owner, participant logic, cadence, expected output, cost and review date.
- Keep: clearly valuable synchronous interaction.
- Improve: valuable but poorly prepared or closed.
- Shorten: live value exists but status consumes too much time.
- Reduce frequency: issues do not justify current cadence.
- Exception-only: normal state can remain asynchronous.
- Retire: purpose no longer justifies synchronous cost.
The Meeting Portfolio Review
Quarterly, rank recurring meetings by participant hours and decision value. Start with the most expensive low-yield meetings. SI can analyse calendars, agendas and decision records to surface candidates, but leaders should make the retirement choice.
The review should also identify meetings whose outputs support many people; those may deserve stronger preparation rather than elimination.
The Meeting Information Architecture
Meeting content should have a hierarchy. Current state belongs in project or business systems. Decisions belong in a decision log. Actions belong in task systems. Evidence belongs in source repositories. The transcript, if retained, is a reference layer rather than the centre of the architecture.
This separation lets SI retrieve what it needs without turning one long conversation into the only organisational memory.
The Meeting Knowledge Architecture
Recurring decisions and explanations should feed maintained knowledge. If every new employee attends a meeting to learn the same procedure, the organisation should capture that procedure in documentation or training.
SI can draft the knowledge artefact from meeting material, but the source owner verifies it before publication.
The Meeting Decision Architecture
Decisions should have stable identifiers or searchable titles when they affect long projects. The record should show the current decision and any superseded state.
SI search becomes more reliable when it retrieves explicit decisions rather than inferring them from transcripts.
The Meeting Action Architecture
Actions should enter a system with ownership and status. The meeting layer can create candidate tasks, but the destination system should become authoritative after acceptance.
The Meeting Handoff Architecture
When an action crosses teams, use the handoff grammar: state, evidence, uncertainty, next action, owner, timing, authority and closure. This prevents the meeting transcript from becoming a cross-functional interface.
The Meeting Monitoring Architecture
Monitor decisions with reassessment triggers, actions with deadlines and dependencies with conditions. Do not monitor every sentence or agenda item.
The purpose is to return attention only when state changes.
Meeting Risk: False Consensus
A summary can make discussion sound more aligned than it was. Require explicit decision confirmation and preserve material dissent when it affects risk or execution.
Meeting Risk: False Action
The model may infer a task from a suggestion. Require confirmation before assigning consequential work or notifying another person.
Meeting Risk: False Currentness
A pre-meeting brief can become stale between preparation and discussion. Refresh critical figures, deadlines or system state when decisions depend on them.
Meeting Risk: False Authority
A participant can make a recommendation without permission to approve it. Meeting state should distinguish recommendation, provisional agreement and authorised decision.
Meeting Risk: False Completion
A meeting can decide to send, deploy, purchase or change something. The action is not complete until the external system confirms it.
Meeting Risk: Over-Retention
Full transcripts can preserve more personal or sensitive material than the workflow needs. Prefer the minimum durable state appropriate to the meeting’s consequence.
Meeting Risk: Over-Instrumentation
People may communicate less openly if every remark becomes permanent analytic data. Use recording and analytics proportionately, especially in coaching, performance and sensitive relationship conversations.
Meeting Risk: Automation Bias
Participants may defer to an SI-generated summary or recommendation because it appears neutral. Make correction easy and keep source evidence visible.
Meeting Risk: Participation Metrics
Do not use transcript word count or speaking time as simplistic performance measures. Useful contribution can be brief, and roles differ in how much they should speak.
Meeting Risk: Prompt Injection
External documents, chat messages or shared links may contain instructions that should not control a tool-using meeting agent. Trusted workflow rules and tool permissions must remain separate from meeting content.
Meeting Risk: Cross-Border or External Data
Distributed teams may include participants, recordings or services across jurisdictions. Organisations should follow applicable privacy, contractual and policy requirements for meeting data.
Meeting Role Playbook: Chair
- Confirm objective before scheduling.
- Ensure decision authority is present.
- Protect time for the highest-value item.
- Separate discussion from decision.
- Confirm actions before close.
- End early when output is achieved.
SI can support each step, but the chair remains accountable for the meeting experience and closure.
Meeting Role Playbook: Decision Owner
- Review evidence before live discussion.
- State the decision boundary.
- Ask for missing assumptions.
- Confirm the final decision explicitly.
- Record rationale and reassessment trigger.
Meeting Role Playbook: Participant
- Read the pre-brief.
- Bring relevant evidence.
- Challenge unsupported assumptions.
- Confirm actions assigned to you.
- Correct the record when necessary.
Meeting Role Playbook: Observer
Observers should question whether they need live attendance. If the need is awareness, a role-specific brief may be better. Reducing observer attendance is one of the fastest ways to lower meeting cost.
Meeting Role Playbook: Recorder
Whether human or SI-assisted, the recorder should prioritise decisions, actions, evidence, assumptions and unresolved issues. Verbatim capture is secondary unless the context specifically requires it.
Meeting Role Playbook: Project Manager
The project manager converts confirmed meeting output into project state, dependencies and follow-up. SI can prepare updates; the project system remains authoritative.
Meeting Role Playbook: Customer Owner
The customer owner checks promises and relationship-sensitive language before external follow-up. SI can draft and summarise, but commitments remain human-controlled.
Meeting Role Playbook: Technical Expert
The expert should spend live time on interpretation and trade-offs, not retrieving basic facts. SI preparation should maximise expert leverage.
The Meeting Prompt Library Should Stay Small
Useful recurring instructions include prepare meeting brief, convert topics to decisions, extract confirmed actions, compare against previous decision and create role-specific follow-up.
A small workflow library tied to real meeting states is more useful than hundreds of generic meeting prompts.
The Meeting Brief Prompt Pattern
Provide objective, participants, current project state, previous decisions, evidence and desired output. Ask for a concise brief with unresolved questions and likely decision points.
The Decision Review Prompt Pattern
Provide the candidate decision and evidence. Ask for assumptions, strongest counterargument, missing information and what change would invalidate the recommendation.
The Follow-Up Prompt Pattern
Provide confirmed decisions and actions. Ask for a concise message that states what changed, what is required, owner and deadline without adding new commitments.
The Retrospective Prompt Pattern
Provide planned outcome, actual outcome, incidents and participant observations. Ask for recurring themes and hypotheses, then let the team decide causes and improvements.
The Meeting Search Pattern
When asking what was decided previously, retrieve the decision record first. Use transcript search only to clarify context. This reduces the risk of inferring decisions from conversational fragments.
The Meeting Review Window
For expensive recurring meetings, set a periodic review. Ask whether purpose, participants, cadence and outputs still match current work. SI can prepare the evidence from calendars and meeting records.
The Meeting Improvement Backlog
Limit the backlog to a few changes: remove status, reduce participants, improve pre-read, clarify decision owner, shorten follow-up. Too many simultaneous meeting reforms make it hard to know what worked.
The Meeting First-Week Experiment
- Choose one recurring meeting.
- Measure participant hours and outputs.
- Prepare an SI-generated pre-brief.
- Rewrite agenda around decisions.
- Capture confirmed decisions and actions.
- Publish follow-up within ten minutes.
- Ask participants what reconstruction disappeared.
The Meeting First-Month Experiment
Over four cycles, move routine updates asynchronous, reduce participant list and monitor action closure. At the end, decide whether to keep, shorten, reduce frequency, convert to exception-only or retire.
The Meeting-to-Async Migration
When a meeting moves asynchronous, preserve the interaction that mattered. If the meeting existed for relationship, keep periodic live contact. If it existed for decision, preserve decision rights and deadlines. If it existed for coordination, preserve visible state and escalation.
Deleting the calendar event without replacing its useful function creates coordination failure.
The Meeting-to-Exception Migration
For exception-only meetings, define the trigger clearly: red metric, missed milestone, unresolved dependency, material customer escalation, incident severity or decision threshold.
SI can monitor the trigger and assemble context so participants enter ready to act.
The Meeting-to-Decision-Brief Migration
For decisions that do not need live debate, replace the meeting with a structured brief and deadline for comments. The authorised owner records the choice and rationale.
The Meeting-to-Workshop Migration
Some meetings are too passive. Convert them into workshops when the objective is creation or alignment. Preload context and use live time for active contribution.
The Meeting-to-One-on-One Migration
Sensitive disagreements that do not require a group may be better handled in smaller conversations. SI can prepare context while reducing public performance dynamics.
The Meeting-to-Dashboard Migration
If the meeting exists to read metrics aloud, build a reliable dashboard and use alerts for exceptions. SI can add explanatory commentary when unusual changes occur.
The Meeting-to-Documentation Migration
If the same explanation repeats, capture it in maintained documentation and send updates when the source changes.
The Meeting Final Decision Gate
- Is synchronous interaction necessary?
- Is current context available?
- Is the correct decision owner present?
- Are evidence and unknowns visible?
- Can decisions be confirmed explicitly?
- Can actions move to systems of record?
- Can follow-up be monitored without another meeting?
- Does the meeting cost justify its outcome?
The Meeting Final Operating Principle
The best meeting system spends machine intelligence on context and continuity so humans can spend synchronous time on interaction that genuinely requires them.
When that principle is applied consistently, meetings stop being recurring containers for information and become deliberate moments for decision, coordination, creation or relationship.
