Three moments in the product lifecycle where AI does real work — and in every one, it sharpens judgment instead of replacing it. Pick one below: each has a worked example and a skill package you can download and run.
Idea → read-ready strategy doc → stress-tested before anyone sees it
Turns messy meeting notes and a rough goal into a strategy document — then runs it past a panel of real operators to find the holes. The value isn't the writing. It's the interrogation up front and the stress test at the end.
How it works
Intake — reads your notes, context, and one-line goal.
Clarify — asks the 4–6 questions that most change the doc, before writing a word.
Draft — one-pager or narrative, and says which it chose and why.
Clarify again — targets its own weakest claims.
Stress test — runs the draft past the panel, then revises.
Step 2 — it interrogates before it writes
Input · raw offsite notes
- onboarding is broken; drop-off spikes at step 3
- competitor shipped a 2-min setup; we're at 11 min
- eng: identity check is the bottleneck, "can't remove"
- ~40% of signups never finish (someone quoted this)
GOAL: get the exec team to fund a 2-quarter rebuild
It refuses to draft — and asks
Who's the actual decision-maker — the CFO, the CPO, the CEO?
If they remember one number, should it be the 11 minutes, the 40%, or the competitor's 2 minutes?
That ~40% — instrumented, or offsite folklore? I'll flag it if it's soft.
What does the reporting-API team lose if these 6 engineers move?
Step 5 — the stress test, and what it actually did
The ask · draft v1
“Approve six engineers for two quarters to rebuild onboarding.”
↓
What the panel said
Good onboarding should sometimes decrease conversion if it improves retention by filtering out unqualified users. Nobody has segmented that 40%.
— Adam Fishman, former CPO, Imperfect Foods
Adding friction to onboarding can increase activation. Removing setup steps that merely look difficult often hurts long-term success.
— Bobby Pinero, CEO, Equals
You're trading a reporting API committed to two enterprise accounts for a modeled activation gain — and pricing neither. That's the whole ballgame for a CFO.
— Shreyas Doshi, former VP Product, Stripe & Twitter
↓
The ask · draft v2
“Approve six weeks to find out whether it's friction or qualification — then a decision gate.”
That's the whole value. Same author, same evidence, one hour later. The panel didn't improve the prose — it changed what he was asking for, and killed an ask that a CFO would have rejected. The doc walked in wanting two quarters and walked out wanting six weeks and a real answer.
Crawl / walk / run — with a working prototype beside each stage
Turns a strategy into one interactive page. A stage selector on top; below it, the strategy on the left and a clickable prototype on the right. Advance the stage and both evolve together — the experience gets richer, the platform capabilities stack up, the value compounds. A room stops debating bullet points and aligns on the actual product.
Example — embedded business payments
Input · a stage outline
strategy: embedded payments for small businesses
stages:
crawl — accept a card payment inside the software
walk — saved methods, autopay, instant payout
run — balance, working capital, spend card
constant: the business never leaves the tool it
already runs its day in
Output · one self-contained page
Click through Crawl → Walk → Run and watch a business go from chasing a
check, to getting paid automatically, to running its money inside the
platform — with the required capabilities stacking as you go.
Is our effort actually going where we said our priorities are?
Pulls historical Jira data and answers two questions leaders lose sleep over: is our time aligned to our stated priorities, and where are the early-warning hotspots. It interviews you before and after seeing the data, so the dashboard serves a decision instead of dumping charts.
How it works
Interview first — what decision, which priorities, what counts as a hotspot.
Pull — epics, points, labels, cycle time, reopens, WIP by team.
Interview again — comes back with what's measurable and what's surprising.
Dashboard — effort vs. intent, hotspots, and a “so what” narrative.
Example
Output · dashboard excerpt
EFFORT vs INTENT (story points)
fraud ........... 41%
mobile parity ... 22%
onboarding ...... 9% ← priority #2, effort #4
unmapped ........ 28%
HOTSPOT
Team Atlas · reopen rate 3.1× median · WIP up 4 wks
So what: either onboarding isn't really priority #2, or it's under-resourced — and there's 28% of unmapped work to reallocate from. The Atlas trend is a signal worth a look, not a verdict.
Deliberate design choice: team- and system-level only, never individuals. It's an early-warning tool to protect delivery and people — not a way to rank anyone.
All three skill packages, the worked example, and the full brief — naming, live-vs-pre-made tradeoffs, talking points, and answers for when senior leaders push back.