Machines that do the work.
Agents, retrieval, analytics and data pipelines — built like factory equipment: one job per station, checked at every step, safe to leave running.
Pick a job below and start the line.
The floor
RECORDED DEMO · example data1 · Choose a job
2 · INPUT SLIPSignup button overlaps the footer on small screens.
Line idle. Load a job and press start.
$ run-log --recorded
waiting for a job…
4 · OUTPUT TRAY
Empty. Start the line to see what comes out.
The machines
Each one has a manual — open a nameplate.
MODEL SF-01
Software factoryAI coding agents for your backlog
InputA ticket from your project toolOutputA tested pull request, waiting for reviewSafetyNever touches main · escalates after retries
Read the manual →
MODEL RT-02
Retrieval engineA chatbot that answers from your documents — with sources
InputYour documents and a questionOutputAn answer that cites its sourcesSafetyEval gate on every change · says "I don't know"
Read the manual →
MODEL AN-03
Analytics deskAsk your database questions in plain English
InputA plain-English question about your dataOutputThe SQL, a chart and the answerSafetyRead-only access · runs on self-hosted models
Read the manual →
MODEL DP-04
Data pipelineInvoice, résumé and document extraction
InputRésumés, PDFs, profiles, web pagesOutputClean, ranked, searchable recordsSafetySchema checks · sampled human audits
Read the manual →
Repair bay. Got an n8n, Make, Zapier or custom AI automation that keeps breaking? Diagnosis and fix.
Book a repair →Clock in.
Need a machine built, or one repaired? Fill in the card and get a suggested setup back.