HK co-create workshop · draft

Capture & readout

From the room to a readout in 48 hours

One Google Sheet, one row per observation, one set of tags.

Start here
  1. Before: import the CSVs below into one Google Sheet that only the team can see.
  2. On the day: note-takers add a row for everything they see or hear, using the note columns and tags.
  3. After: the debrief, then the synthesis, then the readout.

01The sheet

roster.csvnotes.csvbugs.csvideas.csv

One tab per CSV. Add tabs for the paper sheets as you type them up that evening: answers, kcs, scenarios, tokens, plus the survey exports (sign-up, setup, exit).

Names live in one place only. Only the roster tab has names, and only Jeet and Sunny can see it. Everything else uses participant IDs. Nothing from the sheet goes on this site.

02Note columns

A row is one thing: a quote, a behaviour, a bug or an idea. Short rows beat paragraphs.

time · pid10:34 · P07 (or GROUP)
sessionsetup-observed · s1-first-use · s2-ask · s3-concepts · s4-ideas · s5-futures · close
concept1 or 2 (session 3 only)
kindquote (their exact words, in "quotation marks") or behaviour (what they did)
observatione.g. Opened settings looking for "connect Slack", gave up after 90 s
type · severitySee the tags below · 1 minor, 2 slowed them down, 3 blocked them
areaSee the tags below
source · takerlive or scribe · your initials

03Tags

type

pain a problem in their work · delight · confusion · bug (also add it to the bugs tab) · idea · trust (goes to AI Eng) · assist (we had to help)

area

chat · calls · calendar · ask · assistant · connectors · workflows · apps · shell · onboarding · notifications · other

04Debrief · 16:00 · 30 min

  1. Round (15 min): each of us, 2 minutes, no discussion. Top 3 things seen or heard, the most surprising idea, and one thing I believe less now.
  2. Bugs (8 min): P0 blocks first use, P1 hurts daily use, P2 is a rough edge. Give every P0 an owner.
  3. Collection (7 min): account for every sheet, photo and recording, and decide who types up what tonight.

05Synthesis

EveningNote-takersType up the paper sheets and photos. Check Scribe for missed quotes. Tidy the tags.
Day +1Jeet, SunnyGroup the notes into themes for each question. For each theme, record how many people, which groups (A/B/C), and the best 2–3 quotes. Anything from only one person counts as a signal, not a finding. Compare A and B on first run.
Day +2JeetThe readout for the CTO (Matthew Chan), CPO (Aaron Kurzak) and Sunny. The merge brief for Aaron.
+2 weeksJeetA share-back to participants: what we heard, and what we're changing.

Honesty rules: count before concluding. Behaviour beats opinion. Report what went against our expectations as prominently as what supports them. It's 20 people, so the findings are signals, not statistics. Alpha Users are half the room, so give counts by group, not just totals, and don't let their views stand in for new users.

06Templates

Scribe cross-check note-takers
  1. Start Scribe at the start of each breakout session, and say so out loud.
  2. Take live notes as usual. Don't rely on Scribe.
  3. That evening, skim Scribe for your room. Add missed quotes as rows with source = scribe, and fix any you misheard.
  4. Scribe's own mistakes are product feedback too (area calls).
Exit survey Google Form · 5 min · QR code
Exit survey
Before you go (5 minutes) 1. Participant ID [short answer] 2. How likely are you to use ai.com at work next week? [1–5] Why? [short answer] 3. What's the one thing we should build first? [short answer] 4. Finish the sentence: "I'd stop using ___ if ai.com could ___." [short answer] 5. Anything you didn't get to say today? [paragraph] 6. How was the day? [1–5] What would you change? [short answer] ☐ I'm happy to help again: occasional quick tests, no commitment.
Readout template Google Doc · about 4 pages
Readout skeleton
ai.com IC workshop: what we heard [date] · 20 participants (Fresh: 5 · Early-setup: 5 · Alpha: 10) · [n] observations THE SHORT VERSION • Five bullets, one sentence each WHO WAS IN THE ROOM • Function × group, and the range of AI comfort. Who was missing. NEEDS: top pains, each with # people, groups and a quote FUTURES: the six scenarios by want / maybe / no, with the strongest "because" SHELL: keep / cut / steal counts per concept, the Concept 3 patterns → merge brief FIRST RUN: where Group A struggled and what delighted them (with assist counts); how Group B found setup; why Group C keep going or stop NEW IDEAS: the top ideas by dots and tokens, with the problem each one solves PRIORITIES: the token ranking, and the "I'd stop using ___ if" patterns WHAT SURPRISED US (required, never empty) CONCERNS RAISED, for AI Eng (listed as heard) BUGS: P0/P1 count and owners CAVEATS: 20 people, one office, half of them alpha users. These are signals, not statistics.

Workshop kit · Overview · Run of show · Participants ·