Claim Audit
Connect a claim to its decision, evidence source, tested context, exclusions, counterevidence and residual uncertainty, then produce a defensible claim draft.
DO LESS. PROVE MORE.
Complex technical work often fails the credibility test not because teams do too little, but because they claim too much. Do Less. Prove More. is a practical field guide to setting proof boundaries, defining evidence obligations and reducing residual uncertainty before a decision depends on the work.
THE BOOK
This book is for senior engineers, architects, AI/ML practitioners, technical product and program leaders and technology executives who need to make consequential decisions without pretending uncertainty has disappeared.
Its central idea is simple: every meaningful claim creates a proof obligation. The job is not to create the appearance of completeness. The job is to define what is being claimed, test what matters and state what remains unknown.
FIELD TOOLS
All tools run locally in your browser and require no account. They are designed to expose proof obligations and decision boundaries rather than manufacture certainty or hide judgment behind a single score.
Connect a claim to its decision, evidence source, tested context, exclusions, counterevidence and residual uncertainty, then produce a defensible claim draft.
Assess whether the evidence covers the decision-relevant scenarios, failure modes, environment, reproducibility and untested scope.
Frame model version, intended task, evaluation population, metrics, thresholds, comparators, exclusions and what the evaluation actually establishes.
Separate proven conditions, unproven conditions, explicit exclusions and residual uncertainty around a decision.
Review claim precision, traceability, boundary discipline, reproducibility, uncertainty disclosure and decision fit by dimension.
Classify available evidence by strength and determine what additional evidence would materially change the decision.
Capture what remains unknown, consequence if wrong, relevance to the present decision and the next evidence action.
Break a conclusion into linked claims and test whether each inferential jump has an explicit basis.
WORKSPACE · LOCAL FIRST
Save tool states and generated reports on this device, reopen them later and carry a claim into the next proof step without retyping. Nothing in this workspace is sent to a server.
SELECTED WORK ITEM
Saved items include the tool, field state, generated report and timestamp. Reopening restores the inputs so you can continue the work.
MOBILE COMPANION
A mobile companion extends proof discipline into capture, review and cross-device workflows, with evidence attachments, history and deeper challenge patterns.
Coming soon.
Explore Pupfish SoftESSAYS & NOTES
Short essays and field notes extend the book with examples, counterexamples and evolving ideas around claims, proof obligations, trust boundaries and residual uncertainty.
New essays coming soon.
READER LAB
Claim, proof obligation, evidence sufficiency, proof boundary, residual uncertainty and weakest trust boundary.
Worksheets, checklists and working templates. Coming soon.
Video explainers and sessions. Coming soon.
SPEAKING & TEAMS
Executive briefings, team workshops, bulk adoption paths and facilitated claim-audit sessions bring the book's ideas into technical organizations.
Program and media details coming soon.
THE COMPANION BOOK
The Outcome Leader focuses on turning strategy into measurable outcomes through clarity, decision rights, mobilization and evidence.