s1-cover cover Session 1
Mindset · Compounds · Judgement
AI-Native Mindset & Compound Prompting
_demo-s01 · Acme Robotics · 90m · Full Cohort ·
released
_demo-s01 → 6 blocks, 14 slides, 7 trainer notes, 1 exercises, 5 reflects, 2 quizzes, 2 asides
Pre-read
- Anthropic, *Building effective agents* — sections "What are agents?" and "When to use agents"
- Acme's *Whisker brief* (one-pager, in your inbox)
- Optional: PG's race-track-analogy clip (4 min)
- Acme's *Whisker brief* (one-pager, in your inbox)
- Optional: PG's race-track-analogy clip (4 min)
Session goal
Establish the shared mental model the whole bootcamp builds on:
1. **AI-native mindset** — growth is the unlock; fixed is the trap
2. **Compound prompting** beats one-liners — and there's a structure
3. **Context engineering** — Projects and Skills are how Claude knows what it knows
4. **The judgement boundary** — what Claude handles, what *you* sign
1. **AI-native mindset** — growth is the unlock; fixed is the trap
2. **Compound prompting** beats one-liners — and there's a structure
3. **Context engineering** — Projects and Skills are how Claude knows what it knows
4. **The judgement boundary** — what Claude handles, what *you* sign
We open with a frame, not a feature tour. The cohort will spend 14 hours with each other; the first 15 minutes set the air everyone breathes for the rest.
s1-agenda agenda Today
s1-mindset-fixed-vs-growth statement Fixed vs Growth
Fixed mindset asks what AI can't do. Growth mindset asks what you can do because of AI.
The cohort comes in carrying either anxiety (AI replacing them) or skepticism (AI overhyped). Both are fixed-mindset symptoms. Reframe before anything else.
s1-mindset-early statement You're early
You're not behind. You're early — but only if you're moving.
We meet each other through Whisker — Acme's synthetic case product. Not a slide tour. A short brief, then a 5-minute round where everyone shares one Whisker decision they'd want to influence.
s1-whisker-brief split Whisker — the case product
B2B SaaS · 4 customer pilots · 28 active operators Mobile-first on the floor · web companion in the office Open questions on every desk — fe
col 1
- Pilots: Pune, Memphis, Antwerp, Lyon
- Stack: React Native + Node + Postgres + LaunchDarkly
col 2
- Ask: ship the v1 GA feature by Q4
- Constraint: design team is 2 people
The aha block. Same task. Two ways. Two outputs. The cohort sees the lever.
s1-task-setup statement One task. Two ways.
Summarise the attached Whisker PRD into a Slack update for the eng channel. Goal: get a +1 on the proposed launch date.
s1-compound-structure list The compound prompt
The cohort copies this scaffold to their workspace today.
- ROLE — who Claude is for this task
- CONTEXT — what Claude needs that isn't in the input
- INPUT — the material to operate on
- TASK — the single verb
- … 3 more
s1-compound-output split Output — side by side
Same model. Same input. The difference is the scaffolding.
col 1
- One-liner
- Generic. Off-voice. Wrong audience.
col 2
- Compound
- Direct. On-voice. Audience-aware.
Compound prompting is the *what*. Projects and Skills are the *where*. Once you have a structure, you stop typing it every time. You make it Claude's default.
s1-projects statement Claude Projects
A Project is a long-lived context bundle. Upload your DS, your voice, your PRDs once. Every chat in the Project starts already-knowing.
s1-skills statement Claude Skills
A Skill is a packaged invocation. Five-step chains become one command. Your team shares the chain, not the prompt.
This is the block PG should keep notes for — what the cohort surfaces here becomes S2's framing.
s1-outputs-vs-judgement split Outputs vs Judgement
This isn't a line in the sand. It's a moving boundary.
col 1
- Outputs
- First drafts. Lists. Summaries.
col 2
- Judgement
- Decisions. Tradeoffs. The thing you put your name on.
Q&A is the float block. Whatever ran long in Blocks 3 or 4 lands here. Two minutes for S2 setup at the end, non-negotiable.
s1-followup bullets Before Thursday
- Open the Whisker backlog. Skim 10 ideas. Find one dupe.
- Confirm Whisker Aha! sandbox access (test now).
- Pre-read: Backlog signals — one-pager, 15 min.
- Two-question quiz on AI-Native mindset (5 min, in the LMS).
s1-whats-next statement S02 — Backlog Intelligence
600 ideas → 47 dupes → 5 demand themes. In under 3 minutes. Live on Whisker.
s1-thank-you statement Thank you
One thing that landed today — drop it in the reflection prompt below.
After this session
- **Homework:** open the Whisker backlog (linked in chat). Skim 10 ideas. Find one dupe.
- **Setup check:** can you log into the Whisker Aha! sandbox? (test now — fix on Wednesday if blocked)
- **For designers (parallel track):** S2 is PMs-only; designers get S4 on Friday.
- **Pre-read for S2:** Acme's *Backlog signals* one-pager (15 min).
- **Setup check:** can you log into the Whisker Aha! sandbox? (test now — fix on Wednesday if blocked)
- **For designers (parallel track):** S2 is PMs-only; designers get S4 on Friday.
- **Pre-read for S2:** Acme's *Backlog signals* one-pager (15 min).
Add ?token=… to the URL.