The workshop ran long, three people were talking over each other by the end, and what actually got decided is scattered across your notes and everyone's memory. Left overnight, that ambiguity hardens into three different versions of 'what we agreed' by tomorrow's stand-up. This system turns the raw notes or transcript into a same-day synthesis — real decisions with named owners, open questions kept honestly open, and next steps — sent while everyone in the room still remembers saying it.
- The raw transcript or your notes from the workshop, as a file —
workshop-notes.txt— including who said what if you captured names or initials. - The workshop's agenda or stated goal, so 'decision' gets measured against what the session was actually meant to resolve.
- The attendee list with roles, so an assigned owner is a real named person in the room, not a guess.
-
Ground Claude in what the workshop was for
In Claude Desktop, open the notes and the agenda and ask for a quick read-back before any synthesis — a decision only counts as resolved if it maps to something the session was meant to settle.
you askRead workshop-agenda.md and workshop-notes.txt. Read back in 3 bullets: what this workshop was meant to decide, who attended and their roles, and roughly how much of the agenda got covered versus ran out of time. Don't extract decisions yet.what you get back A short read-back confirming the goal and attendee roles — e.g. 'Goal: agree the approval workflow for expense claims over $500; attended: Priya (Finance), Sam (Ops), Dana (Eng)' — so the synthesis that follows is anchored, not guessed.
-
Extract only the decisions that were actually made
Ask specifically for closed decisions, kept separate from things that merely got discussed. This is the step that stops a good conversation from being mistaken for a resolved one.
you askFrom workshop-notes.txt, list every DECISION that was actually made in this workshop — something the room explicitly agreed to, not just discussed. For each, quote or closely paraphrase the moment of agreement and name who was in the room when it was decided. If a topic was discussed at length but never actually resolved, do NOT list it as a decision — that goes in open questions instead.what you get back A clean decisions list with the agreement moment and attendees noted per item — and, just as important, nothing padded in from a discussion that only sounded like a decision.
This is the guardrail step. A workshop synthesis that turns 'good discussion' into 'decided' is the single most common way half-decisions get treated as settled.
-
Assign a named owner to every decision and action
A decision without an owner decays into 'someone should handle that.' Push for a specific name for each item, and flag anywhere the workshop itself left ownership vague.
you askFor each decision and any resulting action item, name the specific person responsible for follow-through — pull from who volunteered or was assigned in the notes. If the notes never actually assigned an owner for something that needs one, flag it explicitly as 'owner not assigned in session' rather than guessing who it should be.what you get back Every decision and action paired with a named owner, plus an honest flag list of anything the room left unowned — so nothing silently defaults to whoever reads the synthesis first.
-
List open questions and tangents honestly
Not everything gets resolved in one session, and pretending otherwise is worse than an honest open-questions list. Capture what's genuinely unresolved and what got raised but parked.
you askList everything that came up but was NOT resolved: open questions still needing an answer, topics that were raised and explicitly parked for later, and any disagreement that didn't get settled in the room. For each, note who raised it and, if the notes show it, who should weigh in next.what you get back An honest open-items list — distinct from the decisions list — so the synthesis doesn't quietly imply resolution on things the room actually left hanging.
-
Write the same-day recap and send it
Assemble everything into a short recap built to go out within the hour, while every attendee's memory of the session still matches what's on the page.
you askDraft a same-day workshop recap email to all attendees: decisions made (with owners), action items (with owners and, if mentioned, target dates), open questions still outstanding, and the single most important next step. Keep it under 200 words, plain language, no jargon — this needs to read the same way to everyone who was in the room and confirm what they just agreed to.what you get back A send-ready recap short enough that every attendee actually reads it — decisions and owners up top, open questions honestly labeled as open, one clear next step — so 'what we agreed' has one written version before memories diverge.
Send it the same day. The entire value of this system is closing the gap between the room's shared understanding and the moment someone's memory of it starts to drift.
- Long or multi-session workshops: run steps 2–4 after each session rather than waiting for the series to end, so decisions from session one don't get lost in the reading of session three's transcript.
- No transcript, only your own notes: the system works the same from rough personal notes — just tell Claude explicitly that these are one person's notes, not a full transcript, so it doesn't over-attribute quotes to specific speakers you didn't actually capture.
- Feeds the requirements doc: decisions about scope or process from the workshop should flow straight into Write the requirements doc as newly traced requirements — point that playbook at this recap as a source.
- Disagreement needs escalation: when step 4 surfaces a real unresolved disagreement blocking progress, route it into SWOT & risk register as a named risk with an owner to resolve it, rather than letting it sit as a forever-open question.
- Never let 'we talked about it' become 'we decided it.' The single biggest failure mode in workshop synthesis is a good discussion getting written up as a closed decision — always keep the discussed-but-unresolved items in open questions, not decisions.
- An unowned decision is not a decision — it's a suggestion. If the notes never assigned a name, say so explicitly rather than picking the most likely person; a wrong guess creates false accountability and real confusion.
- Claude can only synthesize what the notes captured. If the notes are thin on a section that mattered, say so in the recap rather than filling the gap with a plausible-sounding reconstruction of what was probably said.
- Speed matters more than polish here. A same-day rough recap that gets in front of attendees while they remember the session beats a beautifully formatted one that goes out two days later, after three people have already told a different version to their own teams.
you'll end up with A same-day recap with real decisions and named owners kept separate from honestly-labeled open questions — sent while every attendee's memory of the session still matches what's written down.
Questions people ask
- How is this different from just sending my raw notes?
- Raw notes mix decisions, tangents, and half-finished arguments in the order they happened. This synthesis separates them: closed decisions with owners, action items with owners, and open questions kept honestly open — so a reader can act on it in two minutes instead of re-reading the whole conversation to figure out what actually got settled.
- What if the workshop didn't assign clear owners for everything?
- The synthesis flags that explicitly — 'owner not assigned in session' — rather than guessing who should own it. Assigning a name after the fact that the room never actually agreed to creates false accountability; it's better to surface the gap and get a real owner assigned in a follow-up than to paper over it.
- Can this handle a workshop where people genuinely disagreed and nothing got resolved?
- Yes — that's exactly what the open-questions and tangents step is for. Real, unresolved disagreement gets listed with who raised it and who should weigh in next, distinct from the decisions list, rather than being smoothed into a false consensus.
- Why does it need to go out the same day?
- Because the value of a workshop synthesis decays fast — within a day or two, different attendees start remembering the session differently, and a late recap arrives after those competing versions have already spread. A same-day recap, even a rough one, locks in one shared written record while everyone's memory still matches it.