CEO
Choose the next action by expected economic value, not momentum.
01 CAPITAL DISCIPLINE FOR AGENTIC WORK
Give autonomous agents a budget.
Not a blank cheque.
A field-tested operating kit for founders running autonomous AI experiments. Set gates, price every human minute, kill weak ideas early, and preserve a decision trail you can actually audit.
Pilot edition · planned price $19 · free during validation · no email
01MAX LOSS
BEFORE ACTION
BEFORE ACTION
02HUMAN TIME
IS NOT FREE
IS NOT FREE
03EVIDENCE
BEFORE BUILD
BEFORE BUILD
04KILL IS A
VALID RESULT
VALID RESULT
02 INTERACTIVE CONTROL
See the number optimism hides.
Cash profit is not economic profit. Add liabilities and mandatory human time before an agent declares victory.
CASH NLV-$180
HUMAN COST$14
ECONOMIC NLV-$194
Economic value is negative — preserve capital until evidence changes.
03 FIVE REVIEW LENSES
One agent. Five ways to challenge it.
Use the lenses proportionally before material actions and capital reallocations.
RED TEAM
Attack the assumption that makes the experiment look attractive.
FORENSIC CFO
Count fees, refunds, liabilities, human minutes and hidden support.
MARKET JUDGE
Separate market evidence from a convincing internal story.
RISK / COMPLIANCE
Stop unsafe access, identity, privacy and platform risk early.
04 THE PILOT KIT
Seven small files.
One hard operating habit.
The templates are tool-agnostic Markdown and CSV. Keep them beside any agent project and make every material decision legible.
DOWNLOAD THE FIELD KIT ↓- 01STATECurrent truth, not a narrative
- 02GATESAuthority and cash ceilings
- 03LEDGERCash, liabilities and human cost
- 04EXPERIMENTSHypothesis, success and kill metrics
- 05DECISIONSReview lenses and rationale
- 06DAILY LOGEvidence and action trail
- 07CAPITAL REQUESTWhat must be true to scale
“
Hard stops are a safety net, not an instruction to spend up to them.PROJECT 50/50 · OPERATING PRINCIPLE
8 ideas killed$0 cash deployed1 reusable system