THE USE CASE CARD · which AI use cases to pilot first
WAKES WHEN someone asks where to start with AI, which use cases to pick, or how to run a pilot
THE LOOP
discover.py new · the sheet, with the team
score · value, feasibility, risk
the fix for every PREPARE and GUARD
pilot · 30 days, the go bar set first
Write the ledger: ranked · chosen · why
NEVER
Forecast savings the team did not measure
Pilot risky work with nobody reviewing
Set the go bar after the results are in
IT MAKES new — the sheet to fill in with the team; score — PILOT NOW, PREPARE, GUARD, PARK, with every number; pilot — a 30-day plan with the go bar set first.
THE STANDARD the team's numbers, not ours; value is time the work takes now; risky work gets a person first; name the fix; the go bar is set before the pilot; some work should not be automated at all.
PROOF SELFTEST 12/12 · witnessed in Claude 2026-10-09 · Edition 1.
THE SEAL 98 of 100 · LEGENDARY · WITNESSED · Trigger 5 · Machinery 5 · Law 5 · Portability 5 · Proof 5
IT CAN BE WRONG an AI reading a text file: it can be wrong and can invent a number, a rule, a name or a date; check what matters against your own records and the source; the risk is yours. THE-CLAUSES.md rides in the packet and is part of the licence.
THE PUNCH · SHA-256 OF SKILL.MD · 094118C943962D1F434EC0E5
CITE THIS CARDTHE USE CASE CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=use-case-card. Punch 094118C943962D1F434EC0E5. Retrieved 2026-10-10.
What it is.
Ranks a team's candidate AI use cases from its own numbers — hours a week the work takes, how ready the data is, whether success is defined, what a mistake costs, regulation and human review — into PILOT NOW, PREPARE, GUARD and PARK, naming the fix for each, and drafts a 30-day pilot plan with a baseline, an eval set and a go or no-go bar set in advance.
It carries a machine that shows its working: new — the sheet to fill in with the team; score — PILOT NOW, PREPARE, GUARD, PARK, with every number; pilot — a 30-day plan with the go bar set first — 12 checks in its selftest, every command on the record. It will never forecast savings the team did not measure; never pilot risky work with nobody reviewing; never set the go bar after the results are in.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
THE TEAM'S NUMBERS, NOT OURSHours, readiness and risk come from the people who do the work; the machine sorts their judgement and shows every number.
VALUE IS TIME THE WORK TAKES NOWValue is time the work takes now, not a forecast of savings. Savings are measured in the pilot, against a baseline.
RISKY WORK GETS A PERSON FIRSTSerious mistakes with nobody reviewing is GUARD, whatever the value.
NAME THE FIXA PREPARE row says exactly what stands in the way: the data, or an agreed measure of success.
THE GO BAR IS SET BEFORE THE PILOTThe go bar is set before the pilot, never after the results are in.
SOME WORK SHOULD NOT BE AUTOMATED AT ALLSay so plainly; see references/WHEN-NOT-TO.md.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDESreferences/THE-INTERVIEW.md · WHEN-NOT-TO.md
THE MACHINEdiscover.py · three commandsnew — the sheet to fill in with the team; score — PILOT NOW, PREPARE, GUARD, PARK, with every number; pilot — a 30-day plan with the go bar set first. Every command prints its working and refuses a bad input; 12 checks in its selftest, every command on the record.
What it does once your AI has it.
new The sheet to fill in with the team
score PILOT NOW, PREPARE, GUARD, PARK, with every number
pilot A 30-day plan with the go bar set first
On the record The example sheet ranked: refund emails are ready to pilot; contract terms are serious, regulated and unreviewed, so the plan opens by putting a person in the loop first; a weekly update is too small to start with.
What it will never do.
It will never forecast savings the team did not measure; never pilot risky work with nobody reviewing; never set the go bar after the results are in.
The ranking is only as honest as the sheet. It sorts the team's judgement and forecasts nothing. Scores are coarse on purpose: two rows a few points apart are a tie, settled by talking to the team. It is an AI reading a text file: it can be wrong, so check what matters against your own records and the source.
What’s in the buy.
CARD.md — the whole card: who it is, its laws, when it says no, what it carries · SKILL.md — the door your AI reads first · START-HERE.md — the first five minutes, for you
THE-INTERVIEW.md + WHEN-NOT-TO.md — the guides in references/, in the card's own voice
scripts/discover.py — the machine: new · score · pilot, standard library only, selftest 12/12
WITNESS.md + SELFTEST.md — every command on the record, as it printed · MANIFEST.md, every file hashed · AUDIT.md, the twelve checks
MY-LEDGER.md — the card's memory, on your own machine · TIN.md · LICENSE-NOTE.md · THE-CLAUSES.md · HOSTS.md
card.svg + card-machine.svg + card.json · card.html, the card as one offline file · the doors for ChatGPT, Gemini, Cursor, Claude Code and MCP hosts (rules/, system/, mcp/)
PROOF · SELFTEST 12/12 · 3 COMMANDS ON THE RECORD · WITNESSED IN CLAUDE 2026-10-09 · EDITION 1 · PUNCH 094118C943962D1F434EC0E5
ON EVERY DOOR · A CARD FOR YOUR AI, A FILE FOR YOUR COMPUTER
In Claude: turn on code execution under Settings › Capabilities, then Customize › Skills › + › Create skill, and upload the zip as it is — START-HERE.md in the box walks you through it, and tells you how to keep the ledger. In ChatGPT, Gemini and the rest: a Project, a GPT or a Gem with the card’s files attached and the short door in the box pasted as its instructions. The words run anywhere an AI reads Markdown; the machines run wherever the host runs Python. Witnessed in Claude only. The host marks →
Take it.
Nº 830 PROPOSED — LAUNCHING SOON
Sold as is under the term printed. A card instructs an AI you operate; results depend on your model, your files and your judgement. Not medical, legal, tax or financial advice, never the person’s boss, and not a licence to practise. The terms →
One payment. The file is yours to keep and keeps working at its edition; new editions issued in the year after you buy reach you by email as replacement files. Tax is added at checkout where it applies. Nothing auto-renews.
A TERM, NOT A SUBSCRIPTION: NOTHING RENEWS ITSELF AND NOBODY IS CHARGED AGAIN. WHEN A TERM ENDS THE FILES STAY ON YOUR COMPUTER AND STAY READABLE — FREEZE, NOT LOSS.
THE MACHINE · IN THE OPEN · FROM THE WITNESSED RUN OF 2026-10-09
discover.py, every command showing its working.
A small Python program, standard library only. Your AI runs it; you can too. Every block below is pasted from the card’s own WITNESS.md, as printed.
new
the sheet to fill in with the team
score
PILOT NOW, PREPARE, GUARD, PARK, with every number
pilot
a 30-day plan with the go bar set first
selftest
12 checks that prove the machine; it must print 12/12
new · THE SHEET TO FILL IN WITH THE TEAM
$ python3 scripts/discover.py new usecases.csv
sheet written · usecases.csv · three example rows: replace them with the team's own
score · PILOT NOW, PREPARE, GUARD, PARK, WITH EVERY NUMBER
$ python3 scripts/discover.py score usecases.csv
DISCOVERY · 3 use case(s) · usecases.csv
PILOT NOW · Answer refund emails · 40.0 h/wk · value 4 · feasibility 4.4 · risk 1 · priority 17.60 · 40 hours a week, ready, and a mistake is recoverable
GUARD · Pull terms from supplier contracts · 25.0 h/wk · value 4 · feasibility 3.8 · risk 5 · priority 3.04 · risk 5 with nobody reviewing: put a person on every output first
PARK · Weekly team update · 0.5 h/wk · value 1 · feasibility 4.0 · risk 1 · priority 4.00 · 0.5 hours a week: too little to start here
pilot · A 30-DAY PLAN WITH THE GO BAR SET FIRST
$ python3 scripts/discover.py pilot usecases.csv contracts --out pilot.md
pilot plan written · pilot.md · Pull terms from supplier contracts
head -12 pilot.md
$ head -12 pilot.md
# 30-day pilot · Pull terms from supplier contracts
**Team:** Legal ops · **today:** 25.0 hours a week · **ranked:** GUARD (value 4, feasibility 3.8, risk 5)
> **Read first:** this use case is ranked GUARD — risk 5 with nobody reviewing: put a person on every output first. The plan below assumes that is fixed before week 1.
## Before week 1 · agree on success
- One measure everyone signs: already defined — write it at the top of this page
- The bar to go live, set now, not after the results are in.
## Week 1 · the baseline and the eval set
THE NUMBERS, NAMES AND DATES IN THESE RUNS ARE THE STUDIO’S OWN EXAMPLES, INVENTED FOR THE RECORD; NO REAL COMPANY, PERSON OR PROPERTY IS NAMED.
QUESTIONS AT THE TABLE
Asked before the wire.
What does it need from me?
Discover.py new · the sheet, with the team. It asks first for the list of candidate tasks in the team's own words, and for each the hours a week, the state of the data, whether success is defined, and what a mistake costs.
Does it send anything anywhere?
No. The machine is a standard-library Python program that runs on your own computer and sends nothing anywhere; the words run wherever an AI reads Markdown. No server, no account, no key.
What is on the record?
The self-test (12/12) and every command run on real example inputs, pasted into WITNESS.md as they printed; the punch on the face is the SHA-256 of SKILL.md, so a downloaded copy can be checked against this room.
Can it be wrong?
Yes — it is an AI reading a text file. The ranking is only as honest as the sheet. It sorts the team's judgement and forecasts nothing. Scores are coarse on purpose: two rows a few points apart are a tie, settled by talking to the team.
What will it never do?
It will never forecast savings the team did not measure; never pilot risky work with nobody reviewing; never set the go bar after the results are in.
