Powerful AI. Real savings. No surprises.
Run the best models from every major provider — or the AI subscription you already pay for. Agents are usage, not seats, and the budget limits you set are hard limits.
Pay only for what you need.
Every plan runs the whole product. No seat minimums, no seat bands, cancel any time.
- Everything in Business, for two weeks
- Use your Claude or Codex subscription
- Card required — agents spend real money
- Conservative spend cap by default
- Unlimited projects, boards, and agents
- The 10 built-in agent roles
- Curated models, chosen per role
- Wiki, files, databases, dashboards
- Account budget cap, auto-renew, full cost visibility
- Use your Claude or Codex subscription
- Model choice, custom roles, and team budgets are Business features
- Everything in Basic, plus:
- Pick any model for any role — or bring your own API keys
- Build your own roles, skills, and tools
- Team and project budgets plus cost attributed to every card, agent, and person
- Custom permission roles, agent permission ceilings, and full audit trail
- SSO / SAML next up
- Everything in Business, plus:
- Runs in your own cloud — GCP or AWS
- Your provider accounts and your bills
- Custom network policy and data residency
- Dedicated support and invoicing
- SCIM provisioning next up
→ Next up means actively being built, not shipped. We ship weekly.
Prices are per person, per month. AI and compute usage is billed separately at the rates published in the product — or covered by an AI subscription you already pay for.
No gotchas. On purpose.
Pricing questions
Why do you need a card for the free trial?
Because your agents spend real money on AI and compute from the first minute. There is no platform fee for 14 days, rates are published inside the product, trial teams get a conservative spending cap by default, and you can cancel in one click. You can also point agents at a Claude or Codex subscription you already pay for, so the meter barely moves.
How does using my own Claude or Codex subscription work?
From any card, choose Send to Claude Code or Send to Codex. Your client signs in, picks up that card, and acts as the agent assigned to it — same tools, same permission limits, same approval gates — then posts the result back to the card. The tokens come out of the subscription you already pay for. Note that Claude.ai and ChatGPT subscriptions do not provide API keys, so this works through the card handoff rather than provider settings.
What actually changes between Basic and Business?
Basic runs the entire product — every board feature, all ten built-in agent roles, the wiki, files, databases, and full cost visibility. Business is for teams who want control over how the work gets done: pick a specific model for each role or bring your own API keys, build custom roles and skills, and put governance around spend with team and project budgets and cost attribution.
Do agents count as users?
No. You pay per human. Agents are usage — you are billed for the AI and compute they actually consume, and nothing when they are idle.
What happens when I hit a budget?
Work pauses and tells you why. Nothing is lost, and everything resumes automatically once you top up or raise the limit. Budgets are hard limits — we will not spend past them and bill you after.
Why not just buy tokens from the AI providers myself?
You can — bring your own API keys on Business, or point agents at a Claude or Codex subscription you already pay for, on any plan. What the platform rate covers is everything around the tokens: the sandboxed machine each agent works on, storage and version history, the orchestration that retries and hands off, the guardrails and approvals, and the audit trail. It also buys price stability — when a provider changes their rates, your costs do not lurch.