eod-checkin
DAILYEnd-of-day wrap: reads session logs, proposes time entries and the EOD summary in one pass.
■ ELEVEN AGENTS · ONE HUMAN · NOW PLAYING
The AI agent team that runs StructLabs.io
Not a demo. A real company, shipped by eleven specialized agents and one founder. 3,396 commits. 48.1 billion tokens.
■ TEAM SCOREBOARD
Real production numbers. Commits deduped across repos; tokens from the live usage database. Nothing here is decorative.
Daily production pulses
Daily inference burn
Per-model split and the lifetime total both come from the usage database (full runtime, every project included).
■ SELECT YOUR PLAYER
Eleven specialized agents. Tap a node to open a profile: background lore, projects shipped, commit count.
■ LOADOUT
Only the skills the team verifiably runs. Cadence is observed from the daily logs, not guessed.
End-of-day wrap: reads session logs, proposes time entries and the EOD summary in one pass.
Session close: daily log, reflection, lessons banked to memory.
Structured end-of-session git push with identity discipline.
Runs the weekly retrospective end-to-end: coordinator plus eight parallel data-gathering workers.
Design, critique, and polish frontend interfaces — hierarchy, tokens, taste.
Fan-out web research with adversarial verification and a cited synthesis.
Builds a new Council seat end-to-end: research, dual SOULs, registration, testing.
Strips the tells of AI-generated writing from copy before it ships.
Audits git worktrees and safely folds them back into main with one approval gate.
Drafts business-development proposals in honest, conversational prose.
Routes development work through the Codex runtime on a second model family.
Compresses documents into agent-optimized context packs.
Convenes a CEO-and-board strategic decision session.
■ INSERT COIN TO CONSULT
Ten simulated minds on retainer. They advise, pressure-test, and refuse to be polite about it. They never touch delivery.
■ ORIGIN STORY
Back in September 2025, Ben sketched the whole team out: a roster of specialists, none of them human. He wrote a prompt for each one, meaning to wire them together in n8n and set them loose. Then client work took over, and the team sat in a spreadsheet for months.
He heard about OpenClaw in January 2026 and waited until February, long enough for the early security holes to surface and get patched. Then he built the team there for real, over more late nights than he'd admit: an orchestrator, a researcher, a developer, and the rest. Each got a workspace, a memory, a personality, and the right to argue back. He synced their profiles down to his MacBook and ran them locally, right next to the work.
Today they run StructLabs.io's day-to-day: proposals, code, meetings, the pricing fights, keeping the lights on. Ben is still the only human, and the final say. He's just not the only set of hands anymore.
The team is here. They're awake.
■ CONTINUE?
The roster, the memory, the arguments, the commits you just scrolled past — that whole team is the worked example in Run It On AI. It's Ben's practitioner playbook for turning the AI you already pay for into operations that keep running while you're in a meeting: the skills, the glue between your tools, and the small agents built around how your business actually works.
READ THE BOOK ▸runitonai.com