pricing

Pay for the orchestration. Bring your own AI.

Every tier includes the full agent chain. Pricing is per seat — the AI you consume is on your Anthropic bill, never marked up by us. Free is real, not a 14-day countdown.

Free

Try the full pipeline on personal projects.

$0forever
  • 1 seat
  • 1 repo
  • 7-day run history
  • All 22 specialist agents
  • Local runner
  • BYO Anthropic key
  • Community support

Solo

For one developer shipping serious work.

$19/ month
  • 1 seat
  • 5 repos
  • Full run history + replay
  • All 22 specialist agents
  • Jira / Figma / Linear / Slack integrations
  • BYO Anthropic key
  • Email support

Team

For teams that want delivery rigor without the meetings.

$49/ seat / month
  • Unlimited seats + repos
  • All integrations + audit log
  • Org-scoped runner pairing
  • Run history + replay
  • BYO Anthropic key
  • Priority email support
  • SSO (Google + Microsoft)

Enterprise

For regulated industries and large engineering orgs.

Custom
  • Everything in Team
  • Self-host option
  • SAML / OIDC SSO
  • DPA + custom contracts
  • SOC 2 Type II (when available)
  • Dedicated support engineer
  • SLA-backed uptime

faq

What you'd ask before buying

Why bring my own Anthropic key?
Two reasons. First, AI cost stays under your control — pipemason never proxies your traffic and never sees what your code costs. Second, BYO removes us as the middleman, which means we charge for the orchestration layer (the part of pipemason we actually built) and not for tokens we didn't add value to.
Does my source code leave my machine?
No. The runner executes locally. We see metadata (run id, branch, phase, test counts), the iterations.log audit trail, and any artifacts you opt to upload (screenshots, log archives). Source code, environment variables, and secrets stay on your hardware.
What languages and frameworks does pipemason work with?
Any. The runner auto-detects domain tooling from your repo (package.json, Podfile, pubspec, schema.prisma, wrangler.toml, etc.) and infers test/lint/build commands. Specialist agents adapt their idioms to whatever the architects pick — Express vs Fastify, Prisma vs Drizzle, SwiftUI vs Compose.
What if a run fails?
It escalates with full context — the failing trace, a controlled-vocabulary failure class, a suggested recovery action, and the precise file paths involved. You resolve the underlying issue and run pipemason resume. Retries are capped per failure class so the run never silently loops.
How does the Figma integration work?
Connect Figma in Settings (Team and above), paste a Figma frame URL into your run input, and pipemason exports the relevant frames to .pipeline/artifacts/expected/. The screenshot-agent diffs the running UI against those frames at perceptual zero-tolerance.
What does perceptual zero-tolerance mean?
Zero perceptually-different pixels. Antialiasing differences alone don't fail; visible color, spacing, or layout drift does. If you need stricter matching (byte-exact) or looser (structural only), it's a config toggle.
Can I cancel anytime?
Yes — Dashboard → Billing → Manage billing opens the Stripe customer portal where you cancel, update payment method, or change Team seat count. Cancellation takes effect at the end of the current period; remaining days stay yours, no prorated refunds. No annual commits on Solo or Team. Enterprise contracts are negotiated separately.
Is there a free trial on paid tiers?
No — the Free tier itself is the trial. It includes the full 22-agent pipeline, 1 repo, 10 runs per calendar month, and 7 days of run history, and it never expires. When you outgrow those limits, upgrade to Solo or Team. We charge starting that day; we don't ask for a card before you decide.
What counts as a run?
Each 'pipemason start' invocation is one run, regardless of how many phases or iterations the agent chain goes through. A 30-minute run that ships 5 commits and a 45-second run that escalates immediately are both one run. Reruns after a fix (resume) don't count again.
pipemason