Start with structure, then replace every placeholder with information that is true, relevant, and natural for your context. These templates are intentionally skeletal so your final writing does not sound mass-produced.
A reliable AI meeting-notes workflow starts from authorized source material, defines the output structure before summarization, marks uncertainty instead of inventing missing details, verifies decisions and action items against the source, and produces a record that participants can use without mistaking an AI inference for an agreed commitment.
Structured extraction prompt: Using only the source below, return: Decisions | Actions | Named owner | Explicit deadline | Blockers | Open questions | Supporting timestamp/quote. Never infer a decision, owner, or deadline. Write Not stated when absent.
Verification checklist: Item → [ ]. Source line/timestamp → [ ]. Decision or discussion? → [ ]. Owner explicit? → [ ]. Deadline explicit? → [ ]. Human correction → [ ].
Final notes structure: Meeting purpose → [ ]. Decisions → [ ]. Action items (owner + date) → [ ]. Blockers → [ ]. Open questions → [ ]. Next review point → [ ]. Source/record location → [ ].
Privacy gate: Meeting sensitivity → [ ]. Recording consent/policy → [ ]. Approved AI system? → [ ]. Data requiring removal or de-identification → [ ]. Source retained for verification? → [ ]. Human reviewer → [ ].
Constraint decision sheet for AI Meeting Notes Workflow: normal case → [ ]; harder constraint → [ ]; what must stay true → [ ]; what can change → [ ]; evidence/reader/genre check → [ ]; specialist skill to study next → [ ].
Misconception correction audit for AI Meeting Notes Workflow: shortcut or assumption → [ ]; why it fails here → [ ]; corrective move → [ ]; principle that must remain true → [ ]; goal-based next skill → [ ].
Failure-mode and reader-intent audit for AI Meeting Notes Workflow: visible symptom → [ ]; likely underlying problem → [ ]; reader need → [ ]; competing technique considered → [ ]; correction → [ ]; evidence/genre/format constraint → [ ]; final verification → [ ].
Boundary-and-tool-choice audit for AI Meeting Notes Workflow: reader objection → [ ]; actual writing job → [ ]; evidence/genre/format boundary → [ ]; tempting alternative → [ ]; why it is the wrong tool here → [ ]; correct adjustment → [ ]; final verification → [ ].
Evidence/support and mastery ladder for AI Meeting Notes Workflow: minimum acceptable → [ ]; strong → [ ]; expert → [ ]; audience/genre constraint → [ ]; evidence or story support → [ ]; boundary that would require a different technique → [ ].
Certainty and expert-exception audit for AI Meeting Notes Workflow: statement or choice → [ ]; evidence/canon/authority supporting it → [ ]; confidence level → [ ]; uncertainty that must be named → [ ]; expert exception → [ ]; nearest alternative technique → [ ]; why the final choice fits better → [ ].
Purpose-and-verification decision audit for AI Meeting Notes Workflow: real writing job → [ ]; plausible choice A → [ ]; plausible choice B → [ ]; downstream consequence of each → [ ]; facts/evidence/canon to verify → [ ]; reader or genre boundary → [ ]; final purpose-fit reason → [ ].
Publication-readiness audit for AI Meeting Notes Workflow: claim/advice/example → [ ]; strongest available source tier → [ ]; important disagreement or uncertainty → [ ]; what would change the conclusion → [ ]; publish now / qualify / revise / hold → [ ]; final verifier or source of truth → [ ].
Provenance and classification audit for AI Meeting Notes Workflow: draft statement/example → [ ]; classify as fact / interpretation / recommendation / convention / original example → [ ]; source or internal provenance → [ ]; source date / version / checked date → [ ]; attribution needed → [ ]; recency risk → [ ]; final publishable wording → [ ].
Source-record and update audit for AI Meeting Notes Workflow: statement / example → [ ]; external fact, project fact, interpretation, recommendation, convention, or original example → [ ]; controlling first-party / primary source → [ ]; secondary source if used → [ ]; source record / locator → [ ]; date or version checked → [ ]; update trigger → [ ]; publish / qualify / revise / hold → [ ].
Source-ledger and revision-history audit for AI Meeting Notes Workflow: claim / recommendation / example → [ ]; source or project record ID → [ ]; source status active / superseded / corrected / retracted / retired → [ ]; claim-level last verified date/version → [ ]; event that triggered recheck → [ ]; correction / withdrawal action → [ ]; reader-facing disclosure needed? → [ ]; what changed and why → [ ]; prior wording/version retained at → [ ].
Claim/source status audit: Item [claim/source/example] | Status [active / review due / stale / corrected / retracted / superseded] | Controlling source/version [x] | Last verified [date/version] | Stale trigger [x] | Correction severity [0–3] | Reader disclosure [none / note / correction / withdrawal] | Changelog entry [what changed + why].
Dependency and propagation audit: Claim / example / recommendation [x] | Controlling source / project record [x] | Dependency strength [direct / shared / advisory / independent] | Review trigger / reason code [x] | Needs-review scope [local / direct dependents / cluster] | Replacement source requirements [x] | Carry-forward decision [x] | Propagation targets [x] | Reader disclosure / changelog action [x].
Persistent editorial registry audit: Topic slug → [ ]; claim UID/key → [ ]; source UID/key → [ ]; dependency UID + strength → [ ]; source status/version → [ ]; open review UID/reason/priority → [ ]; replacement source if any → [ ]; reviewer + resolution → [ ].