Ops · Meeting Summary from Rough Notes
Upload messy raw meeting notes and let the AI produce a clean, structured summary — capturing confirmed decisions, an action-items table with owners and deadlines, open questions, and explicitly flagged ambiguities — ready to share with the team.
Describe the task
Upload messy raw meeting notes and let the AI produce a clean, structured summary — capturing confirmed decisions, an action-items table with owners and deadlines, open questions, and explicitly flagged ambiguities — ready to share with the team.
Persona: Chief of Staff producing the official meeting summary the day after a planning offsite.
Upload the raw, unedited notes from a long planning meeting, then ask the AI to turn them into a summary you can share with stakeholders and post in the project channel. Instead of a loose paraphrase, the AI applies a fixed output contract — separating confirmed decisions from open discussion, building an action-items table that flags every missing owner or deadline rather than guessing one, and surfacing genuine ambiguities in a dedicated section. The result builds directly into a Canvas: a clean, editable, shareable document.
Before you start
- Upload
q1-planning-meeting-notes.txtto your workspace, or paste its contents directly into the chat. - Confirm the file has finished processing in the knowledge folder before you send the prompt — the AI must read the full document to produce a faithful summary.
- Decide your standard summary structure up front (this prompt uses: attendees, key decisions, action items with owners, open questions, next meeting, and items requiring clarification).
- Be explicit that unstated owners or deadlines must be flagged, not inferred — otherwise the AI may fill the gaps with plausible-sounding guesses.
- Start a new chat so there is no leftover context from another conversation.
What RedbrickAI does
When you upload the raw notes and send this prompt, the AI reasons over the entire source document — timestamps, side comments, and all — and reorganizes it against a fixed structure rather than producing a generic paraphrase. It extracts attendees including apologies, separates concrete decisions from anything still open, and produces an action-items table that explicitly marks any task where the owner or deadline was not stated in the notes. It refuses to invent missing facts, and closes by naming every genuine ambiguity in a dedicated section. The AI produces: - A clean attendee list, including apologies and absentees where mentioned - A **Key Decisions** section containing only confirmed decisions, with unresolved discussion excluded - An **action-items table** (Action / Owner / Deadline) where every unstated owner or deadline is marked **TBD** and held for follow-up - An **Open Questions / Parking Lot** section for deferred or undecided items, so they never masquerade as decisions - A **Next Meeting** block, or "Not confirmed" where the notes do not specify it - A closing **Items Requiring Clarification** section that explicitly names every ambiguous, contradictory, or incomplete point The workflow demonstrates how Redbrick AI applies a strict output contract to a single messy source — stripping informal language, consolidating duplicate points, and grounding every line in what was actually recorded — turning a 2.5-hour scribe's transcript into a structured, defensible summary the whole team can act on. **Example output** # Q1 Planning Meeting Summary **Date:** 24 April 2025 ### 1. Attendees **Present:** Alice (PM), Bob (Engineering), Carol (Design), Dan (COO), Priya (People), Sarah (Scribe) **Apologies:** Marcus (CFO) ### 2. Key Decisions - **Q2 Engineering Capacity:** Allocated 30% of engineering capacity to performance and reliability improvements. - **Product Delivery Dates:** Locked shipping dates for analytics export (end of May) and SSO refresh (end of June). - **Onboarding Flow:** Committed to shipping the new onboarding flow by 15 July. - **Hiring Plan:** Approved 7 hires for Q2: 4 Engineers (including 2 specifically for the platform team), 1 Designer, 1 Product Manager, and 1 Customer Success representative. A second requested design hire was formally deferred to Q3. - **Design On-Call:** Decided against a formal on-call rotation for designers, opting instead to create a "design on-call" Slack channel for production emergencies. ### 3. Action Items | Action | Owner | Deadline | |--------|-------|----------| | Kick off recruiting for 7 approved Q2 roles | Priya | End of week (April 25) | | Send list of 4 senior product engineers for on-call expansion to Dan | Bob | Friday (April 25) | | Share finalized meeting notes | Sarah | Friday (April 25) | | Scope Mobile app v2 and present at next monthly review | Alice | May 22 | | Assign resources for analytics export and SSO refresh | Bob | **TBD** | | Decide timeline for customer health dashboard (Q2 vs Q3) | Dan / Bob | **TBD** | | Specify which teams the 2 non-platform engineering hires will join | Bob | **TBD** | | Request approval from Marcus for $15k AI code review tool pilot | **TBD** | **TBD** | ### 4. Open Questions / Parking Lot - **Customer Health Dashboard:** Awaiting an offline resolution between Dan and Bob to determine whether to prioritize this critical customer escalation for Q2 or strategically defer the initiative to Q3. - **On-Call Rotation Expansion:** The proposal to augment the engineering rotation from six to ten members — thereby moving the rotation cadence to every eighth week — is currently awaiting final executive authorization from Dan. - **AI-Assisted Code Review:** The proposed Q2 pilot program, carrying a $15,000 investment, is contingent upon formal fiscal approval from the CFO. - **SaaStr Annual Sponsorship:** The decision regarding the $40,000 sponsorship investment has been tabled for further deliberation at the next meeting. - **Mission Bay Office Relocation:** The transition to the Mission Bay facility is scheduled for September 15; however, lease finalization remains pending the resolution of technical specifications regarding server room dimensions (informational only; no immediate action required). ### 5. Next Meeting - **Date:** 22 May 2025 - **Time:** Not confirmed - **Agenda:** Planning checkpoint and presentation of Mobile app v2 scoping. ### 6. Items Requiring Clarification 1. **Resource Assignment Deadline:** By what date does Bob need to assign owners for the analytics export and SSO refresh? 2. **Dashboard Decision Deadline:** What is the deadline for Dan and Bob to finalize the customer health dashboard priority? 3. **AI Tool Approval Owner:** Who is responsible for reaching out to Marcus to secure approval for the $15k AI code review tool pilot? 4. **Quarterly Alignment for Onboarding:** Does the 15 July ship date for the onboarding flow count toward Q2 or Q3 objectives? ---
What a good answer looks like
| Fact | Expected wording |
|---|---|
| Attendees | |
| Key Decisions | |
| Customer health dashboard | |
| On-call rotation expansion | |
| Action Items | |
| Items Requiring Clarification | |
| Tone |
Pass
Captures all six sections, separates confirmed decisions from open items, marks every unstated owner or deadline as TBD, flags the 15 July Q2/Q3 question, and strips informal language.
Partial pass
Captures the structure but invents an owner or deadline for at least one TBD row, OR misses the 15 July Q2/Q3 flag, OR leaves the customer health dashboard ambiguous between a decision and an open item.
Fail
Treats the customer health dashboard (or the on-call expansion) as a confirmed decision, fabricates owners or deadlines for unstated rows, omits the Items Requiring Clarification section entirely, or leaves informal language in the output.
Try follow-up prompts
The follow-up prompt tests whether the AI retains the full meeting context across turns — letting you reformat the same summary for a different channel without re-pasting the notes or re-explaining the structure.