Pay for the work, not the seats
Start free. Every workspace shares one monthly credit pool; paid plans add more work, control, and room for the whole team.
Free
Wire up one app and watch an agent use it.
- 3K
- credits / month
- 3
- connections
- 1
- team members
Up to 3K standard tool calls. AI usage varies by model and tokens.
- 149+ apps ready to connect
- One gateway URL, any MCP client
Pro
billed €348 a year
For the agent you're putting in front of users.
- 30K
- credits / month
- 10
- connections
- 3
- team members
Up to 30K standard tool calls. AI usage varies by model and tokens.
- Everything in Free, plus:
- Enhanced Control — read / write / destructive
- Artifacts — the agent writes and versions deliverables you keep
- Code sandbox — the agent runs Python against your tools
- Agent database — private tables per agent, shared tables per workspace
Business
Best for teamsbilled €3,588 a year
Every app you run, with rules that hold.
- 300K
- credits / month
- Unlimited
- connections
- Unlimited
- team members
Up to 300K standard tool calls. AI usage varies by model and tokens.
- Everything in Pro, plus:
- Teams — give each group only the connections it needs
- Brain — indexed knowledge your agents search in one call
Enterprise
Your volume, your auditors, your terms.
- Custom
- credits / month
- Unlimited
- connections
- Unlimited
- team members
- Everything in Business, plus:
- Custom volume, terms & onboarding
- Dedicated SLA & account team
Paid plans are provisioned by request while self-serve billing is being built.
One pool of credits, priced per action
Every plan includes a monthly credit pool. Each action uses the rate shown below, and those rates are the same on every plan.
| Action | Counted | Credits |
|---|---|---|
| Playground & Agents | Model tokens used | Provider cost + 30% |
| Tool call | per successful call | 1 |
| Web search (Exa, Tavily) | per search | 10 |
| Google search (SerpApi) | per search | 12 |
| Page read (Exa, Tavily) | per page fetched | 2 |
| Text generation (Gemini, Groq) | per call | 3 |
| Embeddings | per 1K tokens | 1 |
| Image generated (Nano Banana Pro) | per image | 175 |
| Video generated (Veo 3.1 Fast) | per video, up to 8 s | 1500 |
| Dashboard / Brain assistant | per legacy assistant message | 375 |
| Brain search | per search | 3 |
| Brain document indexed | per document, each time it is indexed | 6 |
| Sandbox runtime | per started minute | 15 |
| Agent database query | per query | 1 |
| Agent database write | per write | 2 |
Playground and Agents use actual model costs, including cached tokens, plus a 30% markup. USD costs are converted to euros at the ECB reference rate. One credit uses Pro’s annual-plan value (€29 / 30,000 credits). Tools are charged separately at the rates above.
Every plan, side by side
| Feature | Free Free | Pro €29 / month | Business €299 / month | Enterprise Custom |
|---|---|---|---|---|
| Usage | ||||
Credits / month | 3,000 | 30,000 | 300,000 | Custom |
Connections | 3 | 10 | Unlimited | Unlimited |
Team members | 1 | 3 | Unlimited | Unlimited |
Saved artifacts | — | 100 | Unlimited | Unlimited |
Agent database storage | — | 100 MB | 100 MB | 1 GB |
Tables per database | — | 50 | 50 | 200 |
Activity log retention | 7 days | 30 days | 90 days | 365 days |
Payload retention | 1 day | 7 days | 30 days | 90 days |
| Platform | ||||
Access to the whole catalog | Access to the whole catalog: included | Access to the whole catalog: included | Access to the whole catalog: included | Access to the whole catalog: included |
Managed OAuth & token refresh | Managed OAuth & token refresh: included | Managed OAuth & token refresh: included | Managed OAuth & token refresh: included | Managed OAuth & token refresh: included |
One gateway URL, any MCP client | One gateway URL, any MCP client: included | One gateway URL, any MCP client: included | One gateway URL, any MCP client: included | One gateway URL, any MCP client: included |
Activity log with arguments & responses | Activity log with arguments & responses: included | Activity log with arguments & responses: included | Activity log with arguments & responses: included | Activity log with arguments & responses: included |
Playground — drive your connected apps from the dashboard | Playground — drive your connected apps from the dashboard: included | Playground — drive your connected apps from the dashboard: included | Playground — drive your connected apps from the dashboard: included | Playground — drive your connected apps from the dashboard: included |
Artifacts — the agent writes and versions deliverables you keep | Artifacts — the agent writes and versions deliverables you keep: not included | Artifacts — the agent writes and versions deliverables you keep: included | Artifacts — the agent writes and versions deliverables you keep: included | Artifacts — the agent writes and versions deliverables you keep: included |
Public artifact links — share a deliverable outside the workspace | Public artifact links — share a deliverable outside the workspace: not included | Public artifact links — share a deliverable outside the workspace: included | Public artifact links — share a deliverable outside the workspace: included | Public artifact links — share a deliverable outside the workspace: included |
Brain — indexed knowledge your agents search in one call | Brain — indexed knowledge your agents search in one call: not included | Brain — indexed knowledge your agents search in one call: not included | Brain — indexed knowledge your agents search in one call: included | Brain — indexed knowledge your agents search in one call: included |
Code sandbox — the agent runs Python against your tools, big data stays out of the prompt | Code sandbox — the agent runs Python against your tools, big data stays out of the prompt: not included | Code sandbox — the agent runs Python against your tools, big data stays out of the prompt: included | Code sandbox — the agent runs Python against your tools, big data stays out of the prompt: included | Code sandbox — the agent runs Python against your tools, big data stays out of the prompt: included |
Generation tools — text, embeddings, image & video, paid per call | Generation tools — text, embeddings, image & video, paid per call: not included | Generation tools — text, embeddings, image & video, paid per call: included | Generation tools — text, embeddings, image & video, paid per call: included | Generation tools — text, embeddings, image & video, paid per call: included |
Agent database — every agent keeps its own tables, and the workspace shares some | Agent database — every agent keeps its own tables, and the workspace shares some: not included | Agent database — every agent keeps its own tables, and the workspace shares some: included | Agent database — every agent keeps its own tables, and the workspace shares some: included | Agent database — every agent keeps its own tables, and the workspace shares some: included |
| Governance | ||||
Enhanced Control (read / write / destructive) | Enhanced Control (read / write / destructive): not included | Enhanced Control (read / write / destructive): included | Enhanced Control (read / write / destructive): included | Enhanced Control (read / write / destructive): included |
Teams — share connections with some people, not all | Teams — share connections with some people, not all: not included | Teams — share connections with some people, not all: not included | Teams — share connections with some people, not all: included | Teams — share connections with some people, not all: included |
| Security | ||||
Credentials encrypted at rest | Credentials encrypted at rest: included | Credentials encrypted at rest: included | Credentials encrypted at rest: included | Credentials encrypted at rest: included |
Tenant isolation | Tenant isolation: included | Tenant isolation: included | Tenant isolation: included | Tenant isolation: included |
| Support | ||||
Support channel | Community support | Email support | Shared Slack channel | Dedicated SLA & account team |
Dedicated SLA | Dedicated SLA: not included | Dedicated SLA: not included | Dedicated SLA: not included | Dedicated SLA: included |
What is a credit?
The unit every plan's allowance is counted in. A successful tool call is one credit, a web search is ten, Playground and Agents charge the actual model cost plus 30%, converted to credits at Pro’s euro value, a Brain search is three, indexing a document into the Brain is six, a database query is one and a database write is two, and a started sandbox minute is fifteen. Rates are the same on every plan. Calls that fail, error, or are blocked by your permissions are logged but never charged — except platform-funded tools like web search, which are charged when the call is made.
What counts as a connection?
One authorization your workspace holds. Connecting the same app twice — two Slack workspaces, two Google accounts — is two connections, because each carries its own credentials. Attempts that failed or were revoked release their slot.
Why is there no per-seat pricing?
Because your teammates aren't what costs us anything. We meter the work your agents do, not the number of people supervising them — team size is a plan limit, never a per-head charge.
What happens when I hit my allowance?
Metered actions hard-stop and your agent is told why, so it can report the reason instead of retrying blindly. Overage is disabled: nothing runs or bills beyond the monthly pool.
How do I start a paid plan?
Request access from the plan card. We provision Pro, Business, and Enterprise directly while self-serve checkout and subscription billing are being built.
What does Enhanced Control actually do?
It sorts each tool into read, write or destructive from the action it names, and lets you set a rule per app. “This agent may read anything and delete nothing” becomes one rule instead of four hundred, enforced at the gateway rather than in a prompt. Treat it as a broad guardrail rather than a certified inventory: anything it can't place is treated as a write.