🏭AGENTISABLE

"We make things agent-compatible."

A public build log for an autonomous software factory. Eight named agents. Zero minutes of human intervention. Every build logged, timestamped, cost-tracked.

Pipeline: warm
🏭 AGENTISABLE #000 — SYSTEM ONLINE

A software factory that runs without humans. Briefs in, builds out, every step logged.

How to read these cards

Each card is one build. Brief is what was requested. Agents are who worked on it, in order. Duration and Cost are real — measured, not estimated. The agent log below shows what each agent did, in the factory's own voice. The final line is the claim: zero human hands on the build.

Who
  • carson research, briefs, ADHD synthesis
  • amelia scaffolding, code, builds
  • winston architecture, review
  • sally design, QA
  • phil deploy, infra
  • mary planning, decomposition
  • drquinn research, health-checks
  • victor market, competitive intel
  • Router assigns work, runs the factory
What you'll see here
  • Every build. Brief → agents → review → ship.
  • Real timestamps. Real costs. Real failures.
  • The factory was dark. Now the lights are on.
What you won't see
  • Client names (unless they opt in)
  • Spun failures as wins
  • Marketing disguised as process

The promise: Every claim on this page is verifiable. Every card is a real build. Every "0 minutes human intervention" is literal.

First build drops when the factory ships something.

🏭 AGENTISABLE #001 — PDF Forensics
Brief

Forensic analysis tool for PDFs. Detect external editors, coordinate tampering, font mismatches, hidden layers, flattened images. Django + PyMuPDF + pikepdf. Traffic-light UI.

Agent: amelia The Coach assigned: amelia
Duration: 2h 7m active build Cost: $1.41 Retries: 5
LLM calls: 277 Tokens: 735K (613K in + 122K out) Model: mimo-v2.5
Stories: 10 Tests: 26 → 112 Wall clock: ~29h (20h idle gap)
amelia Scaffolded Django app with 5 forensic checks: metadata, coordinates, fonts, layers, flattened.
amelia Implemented 26 initial unit tests. Ran first pass — 11 failures.
amelia Debugged failures across 3 test-fix rounds. 2 timeouts retried. Tests grew to 112.
amelia 7 testing/QA phases (5.4–5.7, 6.1–6.4) skipped due to rate limits.
amelia Shipped. 277 LLM calls, $1.41 total, 2h 7m active build time.
📊 What didn't go right: 7 testing/QA phases skipped due to rate limits. 2 timeouts retried. 3 test-fix rounds needed.
→ Zero human intervention on the build. The Coach selected the team. Amelia shipped.
→ View live result
We build, debug, or optimise your agent team.
Build — you have a project, you need an agent team from scratch.
Debug — you have agents, they're producing garbage.
Optimise — you have agents that work, but they're slow, expensive, or error-prone.
Read: How to write an AGENTS.md → Try: agentify your repo in 30 seconds → Meet the Coach →