Resource
Extract Data From Text with AI
Free text in, named fields out. The extraction prompt pattern that turns any unstructured text into consistent, parseable records.
Extract Action Items From Meeting Notes
Decisions actually made, commitments actually given — extracted from fragmentary meeting notes that never label their action items.
Overview
Meeting notes are the messiest extraction source there is: fragmentary, unordered, and full of discussion that sounds like decisions. This setup extracts meeting_date, attendees, decisions, action_items, owners, and next_meeting — with the two field descriptions doing the heavy lifting: decisions means "decisions actually made — not topics discussed", and the reading guidance warns that commitments hide in sentences like "Sara will send the deck". Skip Field keeps the output lean: a meeting with no decisions simply has no decisions key, instead of a row of nulls.
How to use this resource
-
Paste the notes raw
No cleanup needed — the source guidance expects fragments, bullets, and half-sentences.
-
Trust the decisions filter
"Decisions actually made — not topics discussed" is the line that keeps debate out of the record.
-
Mind the Skip Field behavior
Absent fields are omitted entirely — downstream consumers should check key existence, not null.
Why This Works
- Commitment-aware guidance catches the action items notes never label
- The decided-vs-discussed distinction lives in the field description, where the model reads it per-field
- One-entry-per-item list rules make follow-ups countable and assignable
Best for
- Recurring meetings whose notes pile up unprocessed
- Teams that lose action items phrased conversationally
- Follow-up workflows keyed on owners and next-meeting dates
Not for
- Producing meeting minutes or a narrative summary — that's the Structured Summary Prompt
- Judging whether decisions were good ones — extraction reports, it doesn't review
Use cases
- Turning raw meeting notes into a decisions-and-actions record
- Catching commitments phrased as "X will…" instead of labeled tasks
- Listing owners separately so follow-ups have a name attached
FAQ
How do the owners and action_items fields differ when the same person appears in both?
They record different layers of the same commitment. action_items holds each concrete follow-up as a full entry with the owner inline when stated, like 'Send the revised proposal — Jane, by Friday'. owners is a separate deduplicated roster of responsible people — the example lists 'Jane Smith' and 'Marco Ruiz' there, so you can key reminders on a name without re-parsing every task string.
Why does next_meeting sometimes come back as a phrase instead of a YYYY-MM-DD date?
Because next_meeting is defined as a 'Date or trigger', not strictly a calendar date. Its rule applies ISO format (YYYY-MM-DD) only when the date can be determined; otherwise it takes the text as written. A note like 'next sync after the launch' returns that trigger phrase, and the MISSING DATA rule blocks the prompt from inventing a date to force ISO.
How does the extracted JSON keep debate and discussion out of the decisions field?
It leans on the decisions field description — 'Decisions actually made — not topics discussed' — so unresolved back-and-forth is meant to stay out. Extraction reports what the notes state rather than judging whether a decision was sound, so after running the generated prompt in the assistant you use, confirm each listed item was actually settled.
More resources from Extraction Prompt Generator
Resource
Extract Invoice Data with AI
Invoice number, vendor, dates, total, currency — extracted into clean fields with strict no-inference rules, ready for accounts payable.
Resource
Missing Data in AI Extraction — Null, Unknown, or Skip
The most consequential setting in any extraction prompt: what the model does when the field isn't in the text. Four behaviors, and when each is right.
Resources that pair well
Resource
Force JSON Output from AI
Stop getting 'Sure, here is the JSON…' — the output-contract pattern that forces models to return only parseable JSON: schema, example, and a strict rule block.
Resource
Classify Support Tickets with AI
Billing, Technical, Account, How-To, Feature Request — ticket triage with definitions that decide the borderline cases for the model.
Resource
Fix Invalid JSON from AI
The JSON won't parse and you can't see why. Deterministic cause-sniffing — trailing commas, single quotes, unclosed brackets — and the repair prompt that fixes it.
Related tools
Tool
Extraction Prompt Generator
Build prompts that extract defined fields from unstructured text — emails, invoices, tickets, résumés.
Projects that use this resource
Project
Build an AI Meeting Assistant with AI
The full path to an AI meeting assistant — define the use case, turn transcripts into structured notes, extract decisions and action items, classify follow-ups, write a shareable summary, evaluate accuracy, then ready it for the team.
Workflows that use this resource
Workflow
AI Meeting Notes Workflow
Turn a meeting transcript into notes people actually use — a faithful summary, the action items pulled out and assigned, and a clean shareable format.
Guides for this resource
Guide
Turn a Meeting Transcript Into Decisions and Actions
"Summarize this transcript" turns suggestions into decisions, invents deadlines, and drops owners. Here's how to extract decisions, action items, owners, deadlines, and open questions instead — each with evidence and a needs_review flag, and "not stated" for what's missing.