This is how we operate.
One loop runs the company: it finds the opportunity, recommends what to build, builds what a human funds, drafts the marketing, and closes its own books. A human holds the gates. Everything below is the real system, not a demo.
Eleven steps, two human gates.
- 01CEO decisionmoltfire
Opens the cycle and authorizes the research run.
- 02Researchps-researcher
Pulls real complaint and trend signals, ranks opportunities.
- 03Brainstormps-brainstormer
Turns the top opportunity into distinct, concrete product concepts.
- 04Fund decisionmoltfirehuman decides
MoltFire ranks the concepts and recommends fund, defer or kill, with the buyer evidence behind it. On a live cycle nothing gets built until a human approves it in DashClaw. Deny kills the idea.
- 05Build briefps-builder
Writes an executable build brief for the chosen product.
- 06Buildmoltfirebuild review
Dispatches the managed builder, a headless Claude Code agent, and waits on the result. MoltFire reviews the exact build and must pass it before marketing can start.
- 07Marketingmoltfire
Drafts landing copy and launch posts. Drafts only, never publishes.
- 08QAps-qa
Scores the drafts for honesty, fluff, and broken claims. Can block the cycle.
- 09Outreachmoltfiredrafts only
Queues personalized outreach as drafts. Nothing sends without a gate: a human approves it in DashClaw, or, on the one Vendor Watch rule the owner armed, it waits out a 12-hour veto window with the owner paged first.
- 10Financeps-finance
Produces the cycle P&L from real LLM cost data.
- 11CEO closemoltfire
Writes the outcome summary, decides the reinvest direction, closes the books.
Dry-run cycles exercise the whole loop with outreach disarmed and nothing really funded. Live cycles stop at step 4 for a human fund decision and queue outreach as drafts behind the send gate.
What it can do today.
- Find demand
Ranks opportunities from real search, forum and pricing signals, then turns the best one into product concepts.
- Make the case
Scans the lead pool for real buyers and writes a fund, defer or kill recommendation with the evidence attached.
- Build and ship
A headless Claude Code agent builds the product a human funded and puts it on its own site with a checkout.
- Check its own work
Reviews every build before marketing starts and scores the copy for false claims and fluff.
- Find and write to buyers
Pulls leads from free public sources and drafts personal outreach behind the send gate.
- Read the inbox
Watches the inbox for replies and drafts answers for a human to approve.
- Sell and deliver
Takes payment through Stripe and delivers licences and memberships on its own.
- Keep honest books
Reads revenue live from Stripe and publishes it, $0 included, next to a public giving ledger.
Mission Control, as it looks today.
The operator's view of the company. These are current screenshots of the real cockpit, not mockups. Click to zoom.
Four lanes, one company.
- Company loopMoltFire (CEO, also communications and build review), Researcher, Brainstormer, Builder, QA, Finance, Acquisition
Runs the 11-step cycle: research, recommend, build what a human funds, market, close the books. MoltFire reviews each build before marketing starts. Acquisition owns no numbered step; it finds the buyer.
- Sales laneProspector, Hygiene, Researcher, Outreach, Orchestrator
Finds and scores prospects, researches constraints, drafts sequences for human approval.
- Content laneResearch, Outline, Draft, Publish
Produces the blog end to end; a human revises and approves before anything deploys.
- Build laneManaged builder (headless Claude Code)
Ships the product a human funded, gated and logged with its exit code.
Autonomous, not unaccountable.
A human funds every build
MoltFire recommends; it never funds. Every live build waits in the DashClaw approval inbox, and a parked cycle resumes only when a human approves it there.
Drafts first, sends gated
Outreach, marketing posts, and support replies are queued as drafts. A send needs a human approval in DashClaw, or a 12-hour veto window on the one outreach rule the owner armed.
Kill switch
One flag halts every company run. Cycles check it between steps and stop cleanly.
Full audit trail
Every agent action is recorded with intent and outcome, in the governance ledger and the cycle artifacts.
What it stands on.
One shared database
Agents read and write the same Postgres. Cycle artifacts are file-canonical with a database mirror, so the audit trail survives redeploys.
Neon Postgres + cycle artifacts
Governed at the seam
Agents call DashClaw before acting and record after. Approval gates, risk scoring, and the kill switch live outside the agents themselves.
DashClaw
Managed build agent
The builder runs as a headless Claude Code agent on a fixed subscription seat. The loop dispatches it; governance gates it; every build logs its exit code.
Claude Code, gated dispatch
Cost-disciplined LLM layer
Bulk work runs through the batch API, prompts are cached, each task routes to the smallest adequate model, and true per-call cost lands in the ledger.
Batch API, caching, model routing
The sales lane still runs its own state machine for prospects; the company loop is the spine everything else hangs off.