s6-B1-1 pb-cover Combine your tools · pick your capstone PRD
M1 done. M2 done. Tonight we bridge to M3.
Cohort catchup · combine your tools · pick your capstone PRD
s06 · Betterworks · 90m · Full Cohort ·
released
s06 → 5 blocks, 25 slides, 19 trainer notes, 7 exercises, 1 reflects, 0 quizzes, 0 asides
Pre-read
- Bring the **Skill you built in S03 / S05** — invocable in claude.ai (PMs: your M1 skill · Designers: your BW design critic)
- Have your **claude.ai Design Project** open (Figma connector active, DS files loaded)
- Skim the **S06 gdoc** for the PRD shell template (link in your cohort home)
- One **BW use case you'd love to see prototyped** — half-formed is fine; the room refines together
No essays. Show up with your Skill working + one half-formed BW idea.
- Have your **claude.ai Design Project** open (Figma connector active, DS files loaded)
- Skim the **S06 gdoc** for the PRD shell template (link in your cohort home)
- One **BW use case you'd love to see prototyped** — half-formed is fine; the room refines together
No essays. Show up with your Skill working + one half-formed BW idea.
Session goal
By end-of-session:
1. **Cohort is binding ceremony done** — everyone remembers what we covered, what's new, what we can do
2. **M1 + M2 + M3 framing is clear** — M1 done (backlog intelligence), M2 done (DS + design Skill), M3 ahead (mobile prototype + backend connection)
3. **Mid-survey complete in the room** — no homework
4. **Everyone has tried the combine** — Skill running inside a Project chat at least once
5. **Each of 3 groups has a PRD shell** in the S06 gdoc, ready for S07 to prototype against
1. **Cohort is binding ceremony done** — everyone remembers what we covered, what's new, what we can do
2. **M1 + M2 + M3 framing is clear** — M1 done (backlog intelligence), M2 done (DS + design Skill), M3 ahead (mobile prototype + backend connection)
3. **Mid-survey complete in the room** — no homework
4. **Everyone has tried the combine** — Skill running inside a Project chat at least once
5. **Each of 3 groups has a PRD shell** in the S06 gdoc, ready for S07 to prototype against
Cohort catchup · what landed in S01 · S02–03 · S04–05
19 itemsThe recap is the binding ceremony. Three phase beats. Each lands with a quiet anonymous "what stood out" footer — PG names the cohort members live. Two short demos break up the recap with real cohort artifacts.
s6-B1-2 list From S01 — what we landed
The four bets the programme built on.
- Default to AI — Claude tab opens before everything else. Drafts start there; decisions end with you.
- Pick the model on purpose, not on habit.
- Structure beats model — compound prompting is the lever.
- Outputs are artifacts. Judgement is the commitment.
s6-B1-3 list From S02–S03 — M1 · Backlog Intelligence
The 5-step chain · packaged as a skill · running in your terminal.
- The chain — extract · dedupe · map · cluster · report
- Compound → skill — the chain packaged as one command (Eden's
/marked-for-death) - Claude Code in project folders —
cd,/init,/usage, plan mode - Cost as a lever — 9¢ Sonnet vs 45¢ Opus · pick on purpose
s6-B1-4 activation Cohort catchup — who'll show their M1 skill?
One PM walks through what they shipped in S03. Show the skill running on a real BW slice.
s6-B1-5 list From S04–S05 — M2 · Design system in Claude
Your DS as the contract · your Critic Skill packaged · personas as the lens.
- DS as guardrail — tokens are the contract Claude refuses against
- Design Project + connector — Figma connector · DS files loaded · the critic that knows your tokens
- Persona-lens critique — same screen, different persona, different P0
- Your Skill is yours — packaged in claude.ai · invocable across all conversations
s6-B1-6 activation Cohort catchup — who'll show their design Skill?
One designer walks through what they built in S05. Show the Skill running inside a Project chat.
s6-B1-7 activation Anyone else have a takeaway?
Two reacts max · drop in chat or unmute.
s6-B1-8 custom (no title)
M1–M2–M3 orientation · where this is heading · mid-survey
11 itemsOrient to the arc · land the destination (M3 roadmap · the cert · the team-energy voice-over) · then mid-survey in the room. No homework.
s6-B2-1 section M1 → M2 → M3
Two done. One ahead. Capstone weaves them together.
s6-B2-2 list Where we are · where we're going
M1 + M2 done. M3 is the bridge to capstone showcase.
- M1 ✓ Backlog Intelligence — the chain · the skill · run on real data
- M2 ✓ Design system in Claude — DS Project · Critic Skill · persona-lens
- M3 → Mobile prototype + backend — S07 prototypes in Figma Make · build break refines · S10 wires to real data via MCP · S11 ships to leadership
s6-B2-3 list M3 — where this is all heading
S07 polishes the PRD · break builds · S08 clinics · S09 wires · S10 ships.
- S07 (Thu) — polish the PRD · write · critique · lock with the wiring tier committed
- Build break (26 Jun → 12 Jul) — build the prototype against your locked PRD · 4 hr coaching budget
- S08 (Mon 13 Jul) — capstone clinic · M1 + M2 walk per group · cohort hot-seat critique
- S09 (Tue 14 Jul) — M3 wiring to your committed tier (sheet / API) · two dry-run rounds
- … 1 more
s6-B2-4 custom (no title)
s6-B2-5 activation Mid-survey
How's the programme landing for you? Honest signal · anonymous OK.
PRD critic · into your shared Claude Project
9 itemsEach group dumps its raw PRD-writing prompts into the S06 gdoc, runs them through a **reviewer** + **fuser** pair, and ends with one fused group PRD skill they upload to their shared Claude Project. Designers, in parallel, grab the **UX critique seed** for Gauri's refinement — that side lands in S07.
s6-B3-1 section Your raw inputs · one fused group PRD skill.
Dump · review · fuse · upload. Shared Claude Project per group.
s6-B3-2 split Two master prompts · one fused skill
PMs review and fuse. Designers grab a seed for Gauri.
col 1 · PM master prompts (S06 gdoc · Tab 2)
Reviewer — audits each PM's raw PRD prompt against a 7-item checklist.
Construct
col 2 · Designer seed (S06 gdoc · Tab 3)
UX critique seed — persona-lens · DS-token grounded · refusal-bearing.
Gauri ref
s6-B3-3 list The 4 steps — open the S06 gdoc and run them in order
~18 min total · Tabs 1 → 2 → 4 (PM) · Tab 3 in parallel for designers.
- 1 · Dump (Tab 1) — each PM pastes their existing PRD-writing prompt as markdown · 1 section per PM
- 2 · Review (Tab 2) — run the
prd-prompt-reviewermaster prompt on each raw input · paste each coverage map back under its PM's section - 3 · Fuse (Tab 2) — run the
prd-prompt-fusermaster prompt with all raw prompts + coverage maps · produces one fused group PRD skill - 4 · Upload (Tab 4) — paste the fused skill into Tab 4 · upload to your group's shared Claude Project as a skill
s6-B3-4 activation Run the 4 steps · share your fused skill name
One fused skill per group · paste in Tab 4 · drop the name + one notable add in chat.
PRD selection · three groups · three BW use cases
13 itemsThree breakout groups. Each picks a real BW use case. Drafts a PRD shell in the S06 gdoc. S07 prototypes against it.
s6-B4-1 section Three groups · three BW use cases · three PRDs
One PRD shell per group · S07 prototypes against this.
s6-B4-2 list Feature constraints — choose a PRD that can ship
Five must-haves. Filter your idea before you write it.
- Real — a workflow you or your team actually does (not a hypothetical)
- Mobile — fits an under-30-second moment, not a desk-bound flow
- Wirable — has data that already lives somewhere you can read (BW API · Google Sheet · existing dataset)
- Small — one screen, one read, one write at most
- … 1 more
s6-B4-3 split Wiring-tier ladder
Where will your prototype get its data? Pick a tier at PRD time.
col 1 · T2 · Google Sheet backend (default)
Smart default.
You own a sheet that holds the data your screen reads.
Sheets API
col 2 · T1 · Real BW API (stretch)
Gold standard. Only if confirmed by S06/S07.
Needs swagger docs OR an existing M
s6-B4-4 list What each group ships in the gdoc
Tight 6-field shell · clarity over completeness.
- Problem — what BW pain are we solving? (one sentence)
- Target persona — which BW persona, and why this one?
- Smallest M2 proof — one screen, one flow that proves the spec
- DS constraints — which tokens · components · refusal rules apply
- … 2 more
s6-B4-5 activation Breakout · 3 groups · S06 gdoc
Form groups by who's nearest the same BW pain. One section per group.
s6-B4-6 activation Walk your group's PRD
Read your gdoc section aloud. One cross-observation from another group.
S07 preview · close
5 itemsLand the handoff to Thursday. The M3 destination, the cert, the team-energy moment all happened in B2 — B5 is just the concrete pivot to S07.
s6-B5-1 recap Next time — S07 · Polish your PRD
Thu 25 Jun · 19:00 IST · Full Cohort
- Bring your group's PRD shell (from tonight's gdoc) · with data source + wiring tier named
- Veronica's PRD-writing skill develops the shell · the **PRD critic** you set up tonight hammers it for ship rigor
- We seed the **UX critic** prompt in S07 — both critics live in your shared Claude Project from then on
- By end of class — three locked PRDs, one per group · ready for the build break
s6-B5-2 pb-thankyou (no title)
After this session
- **Brief your group members** who weren't in the room — they need to walk into S07 aligned on your PRD shell
- **Your shared Claude Project stays the team's home** — PRD critic seeded tonight; UX critic lands in S07; both serve the build break
- **For S07 (Thu 25 Jun):** bring your group's PRD shell URL + your design Skill (working) + your Figma Make access + access to the shared Claude Project you set up tonight
- **Failure-mode recovery (trainer-side):**
- **B3 setup runs long** (Claude Projects UI snags) → drop the share to 60 sec; push B4 breakouts by ~5 min — B4 is the elastic block today. If B4 still slips, the group walks (slide 23 activation) is what rolls over into S07.
- **B4 breakouts run short** for one group (a PRD that nobody wants to anchor) → PG seeds: *"here's one I think BW would care about — [X]. Anyone want to take it?"* Don't let a group sit empty.
- **Mid-survey completion < 60%** at end of B2 → drop a final 90-sec reminder at the start of B5, keep the link visible.
- **If demos run long in B1** → cap "Anyone else?" prompt to one react max, claw back 1 min.
---
## Open items (for polish pass / PG review)
1. **Mid-survey URL** — `[?CONFIRM]` placeholder in B2 activation slide. The check-in banner reads from `check_in_mid_open_at: 2026-06-24` already in the cohort YAML; verify the banner renders on cohort home Wednesday morning.
2. **Cohort catchup demo names** — left blank in slides intentionally. PG calls names at delivery (Paul for M1, Gauri for M2 based on session notes, but PG flexes by who's in the room).
3. **Praise footers** — three footers reference specific cohort moments without naming. PG fills names live (Veronica + wishlist depth for S01 · Paul's Confluence reuse for S02–S03 · Gauri's persona doc + Mahesh's Vue bridge for S04–S05).
4. **Team-energy voice-over** (B2 statement, was B5) — the script is intentionally light because PG's specific recollection lands better than my draft. Anchor to 1–2 specific contrasts from past trainings.
5. **Activation count** — 6 activations across B1–B5 (2× cohort catchup demos · "anyone else" prompt · mid-survey link · PRD critic drill · breakout · group walks) · 4 surfaces (screen-share · chat · link · gdoc). Passes §17 floor.
6. **Group formation timing** — assumed ~12 in the room, 3 groups of 4. If attendance differs significantly, PG flexes group count at runtime (2 groups of 6 OR 4 groups of 3).
7. **PRD critic prompt** — full version at `prompts/s06/lean-pm-kit/prd-critic.md`. Seed into the **S06 gdoc** under a "PRD critic — install into your shared Claude Project" section so the cohort can copy live.
8. **UX critic prompt** — deferred to S07 (when prototypes exist). Author `prompts/s07/.../ux-critic.md` during the S07 iteration and seed into the S07 gdoc + the same shared Claude Project.
- **Your shared Claude Project stays the team's home** — PRD critic seeded tonight; UX critic lands in S07; both serve the build break
- **For S07 (Thu 25 Jun):** bring your group's PRD shell URL + your design Skill (working) + your Figma Make access + access to the shared Claude Project you set up tonight
- **Failure-mode recovery (trainer-side):**
- **B3 setup runs long** (Claude Projects UI snags) → drop the share to 60 sec; push B4 breakouts by ~5 min — B4 is the elastic block today. If B4 still slips, the group walks (slide 23 activation) is what rolls over into S07.
- **B4 breakouts run short** for one group (a PRD that nobody wants to anchor) → PG seeds: *"here's one I think BW would care about — [X]. Anyone want to take it?"* Don't let a group sit empty.
- **Mid-survey completion < 60%** at end of B2 → drop a final 90-sec reminder at the start of B5, keep the link visible.
- **If demos run long in B1** → cap "Anyone else?" prompt to one react max, claw back 1 min.
---
## Open items (for polish pass / PG review)
1. **Mid-survey URL** — `[?CONFIRM]` placeholder in B2 activation slide. The check-in banner reads from `check_in_mid_open_at: 2026-06-24` already in the cohort YAML; verify the banner renders on cohort home Wednesday morning.
2. **Cohort catchup demo names** — left blank in slides intentionally. PG calls names at delivery (Paul for M1, Gauri for M2 based on session notes, but PG flexes by who's in the room).
3. **Praise footers** — three footers reference specific cohort moments without naming. PG fills names live (Veronica + wishlist depth for S01 · Paul's Confluence reuse for S02–S03 · Gauri's persona doc + Mahesh's Vue bridge for S04–S05).
4. **Team-energy voice-over** (B2 statement, was B5) — the script is intentionally light because PG's specific recollection lands better than my draft. Anchor to 1–2 specific contrasts from past trainings.
5. **Activation count** — 6 activations across B1–B5 (2× cohort catchup demos · "anyone else" prompt · mid-survey link · PRD critic drill · breakout · group walks) · 4 surfaces (screen-share · chat · link · gdoc). Passes §17 floor.
6. **Group formation timing** — assumed ~12 in the room, 3 groups of 4. If attendance differs significantly, PG flexes group count at runtime (2 groups of 6 OR 4 groups of 3).
7. **PRD critic prompt** — full version at `prompts/s06/lean-pm-kit/prd-critic.md`. Seed into the **S06 gdoc** under a "PRD critic — install into your shared Claude Project" section so the cohort can copy live.
8. **UX critic prompt** — deferred to S07 (when prototypes exist). Author `prompts/s07/.../ux-critic.md` during the S07 iteration and seed into the S07 gdoc + the same shared Claude Project.
Add ?token=… to the URL.