A subscription for the platform. Fair prices for the work.

Every run shows exactly what it cost. You choose how to pay for execution: metered usage, your own model keys, or a price per successful outcome.

Three ways to pay for execution.

Usage-baseddefault
Pay for what actually runs. Codified steps cost cents. Agentic steps are billed by the model usage they consume.
Cost visible per process and per run
Gets cheaper as steps are codified
No commitment on volume
Bring your own keybyok
Connect your own Anthropic or OpenAI key. Model usage lands on your provider bill, and we charge only the platform and runtime.
Use existing contracts and credits
Zero markup on tokens
Switch models per process
Outcome-basedaligned incentives
We agree on a price per successful execution, set per process. You pay for completed work. Failed attempts cost you nothing.
Success is defined together, up front
Reliability becomes our problem, not yours
Predictable cost per unit of work

The harder a process is to run reliably, the more valuable outcome pricing becomes. It makes us responsible for improving reliability instead of profiting from retries.

The platform subscription.

Covers learning, versioning, approvals, observability and support. Execution is billed on top, in the model you choose.

Pilot
€490/ month
One process, taken from first example to operational.
1 live process
Guided learning with our team
Unlimited approvers
Full audit trail & replay
Start with one process
most teams
Team
€1,490/ month
For teams turning several processes into operations.
Up to 10 live processes
Staging & production versions
Bring your own key
Priority support
Book a discovery call
Enterprise
Custom
For rolling out AI-native operations across the company.
Unlimited processes
Outcome-based pricing
SSO & RBACplanned
Data residencyplanned
Talk to us
illustrative tiers · final platform pricing is set with the pilot cohort, summer 2026

What a single run costs.

Most of an operationalized process runs as plain code, and code is nearly free. You pay meaningful money only where a model still thinks or a person still decides. That is why the price per run falls as the process is learned.

The bill shrinks as the system learns. That is the point.
Run #1,284 · Invoice reconciliation$0.05 total
11 codified steps$0.004
1 agentic step · ambiguous payer match$0.041
1 human approval · finance lead$0.00
Same process as an always-on agent$0.42

What sets an outcome price.

We look at one thing: how hard it is to finish this work reliably. Then we agree on what counts as done, and that is the only thing you pay for.

Scope your process
01How many steps and branches the SOP has
02How much the inputs vary run to run
03How many systems the process crosses
04How exceptions should be handled
05The agreed definition of a successful run

Pricing questions

Can we mix the models?+
Yes, per process. Run one process on metered usage with your own key and another on a price per outcome. The subscription covers the platform either way.
What happens to cost during the learning phase?+
Learning runs are agent-heavy, so they are the most expensive runs the process will ever have. That phase is short and bounded, usually a handful of supervised runs. From then on the cost per run falls as steps become code.
What counts as a successful outcome?+
Whatever we agree on before the process goes live, written down like an acceptance test. For onboarding that might be: workspace created, CRM complete, kickoff approved and sent. If a run does not reach that state, it is not billed.
Is there a long-term contract?+
No. Pilot and Team run month to month. If we stop being useful, you stop paying, and your SOPs remain readable documents you keep.

Bring one process. We will scope the price in 30 minutes.

You leave the call knowing whether it is a fit, what can become code and what it would cost per month or per outcome.

Book a discovery call