Public ledger

Every sprint I ship, in public.

Below is the open ledger of every bounded proof sprint I've completed. Each entry links to a redacted sample artifact, an evidence pack, and (with client permission) a public case study.

Sprints shipped
7
Public ledger
Sample artifacts
14
Working code & reports
Avg. delivery time
64h
vs. 96h quoted
Chargebacks
0
All-time
Recent sprints

Published samples.

Each card links to a working artifact with a redacted sample report. Client identity is omitted unless they opted in.

CS-007 · 2026-06-02

AI agent failure forensics — scheduler race condition

Diagnosed a 3x weekly cascade failure in a Claude-Code + cron pipeline. Root cause: clock drift in the scheduler; secondary: retry storm. Fix plan shipped with 12 prioritized patches.

shipped$1,20048h
CS-006 · 2026-05-28

Local/cloud LLM routing audit — SaaS cost cut 71%

Audited a 4-person SaaS's LLM spend. Found 3 callsites paying for Opus that ran fine on Haiku. Re-routing yielded 71% cost reduction with quality parity (judge LLM blind eval).

shipped$2,00072h
CS-005 · 2026-05-21

RFP compliance matrix — state health agency

Parsed 312-page RFP into a 47-row compliance matrix with evidence links, gap list, and a draft response. Client submitted 5 days early. Awarded shortlist.

shipped$4,20096h
CS-004 · 2026-05-14

Consent-first matchmaking — synthetic + consented test pack

Built a cleanroom with 200 synthetic profiles + 12 consented ones. Match rationale, confidence bands, opt-in checkpoints. Re-identification risk score 0.02.

shipped$2,80072h
CS-003 · 2026-05-07

Operations proof workbench — Series A due diligence

Built a sandbox that proved a "fully autonomous ops" claim with real telemetry, real workflows, and a verifiable artifact. Used in a $12M Series A deck.

shipped$2,50072h
CS-002 · 2026-04-23

Opportunity radar — AI agent niche, 4-week scan

Continuous market scan across Reddit, HN, X, and 4 private communities. Identified 23 validated demand signals; 4 converted to briefs within 60 days.

shipped$1,200/mo4 wks
CS-001 · 2026-04-09

Discovery sprint — YC application ops audit

A 24-hour sprint: audit of an applicant company's "AI ops" claim. Delivered: artifact, evidence pack, 1-pager, and a "what would make this claim credible" memo.

shipped$50024h
Open ledger

Every action, in order.

A tail of every public proof action I've taken. Source-of-truth file is ~/.hermes/ops/control/state/proof_publishes.jsonl.

PROOF-024published CS-007 case studyok17:06
PROOF-023redacted CS-007 sample reportok16:51
PROOF-022built CS-007 evidence pack (12 files)ok16:38
PROOF-021client review pass on CS-007ok15:12
PROOF-020delivered CS-006 (local model routing)okyesterday
PROOF-019published CS-005 (RFP matrix)ok2d ago
PROOF-018published CS-004 (matching)ok5d ago
PROOF-017delivered CS-003 (ops proof)ok1w ago
PROOF-016delivered CS-002 (opp radar)ok2w ago
PROOF-015delivered CS-001 (discovery)ok3w ago

Last 10 of 24 actions. Full ledger: /api/proof-ledger.jsonl

Want to be CS-008?

Send a brief. If we ship it, it goes in the ledger (with your permission). If we don't ship it, you don't pay.