PRICING
Put independent policy, trusted context, human approval, signed grants, and decision evidence in the execution path — across agent frameworks, clouds, and custom runtimes.
Production plans from €750/month, billed annually.
WHAT YOU BUY
Plans scale with the governed scope: workspaces, active agents, decision capacity, and evidence history — not the number of administrators or approvers.
Identity, policy, trusted evidence, and approval remain outside the agent being governed.
Each plan includes a clear monthly allowance for accepted governed action requests.
Signed grants and decision history show what was permitted, refused, or held for a person.
Start in a sandbox, govern a first production application, or carry one approach across several teams.
DEVELOPER
Free
For development and evaluation
Build and test an Agent Control Room integration before production.
No production service commitment.
PRODUCTION
€750/month
Billed annually at €9,000
Govern the first live agent application with the complete enforcement loop.
Annual production plan with assisted onboarding.
SCALE
€1,500/month
Billed annually at €18,000
Apply one governance approach across several teams, environments, or applications.
For established programmes spanning several governed environments.
ENTERPRISE
From €3,500/month
Billed annually from €42,000
For high-consequence, regulated, or large-scale agent deployments.
Final scope and availability are agreed before commitment.
Prices exclude VAT and other applicable taxes unless stated otherwise in the order form.
OPTIONAL COMPANION SERVICE
Agent Control Room enforces deterministic rules, trusted context, human approval, and signed grants on its own. Organisations that also subscribe to Complyee Enterprise can pair a workspace with the Complyee Policy Library for an independent reading of written corporate policy before selected actions proceed.
Records a supported, flagged, or unclear policy reading with the decision and any approval request, without changing the Agent Control Room outcome.
Preserves an Agent Control Room allow only when the reading supports it; otherwise the action is held for a person or refused according to the configured fallback.
Complyee Policy Library never grants permission and cannot override an Agent Control Room refusal or human-approval requirement. Pairing is configured per action and is optional.
Agent Control Room plans and governed-decision allowances remain unchanged. Complyee Policy Library access requires Complyee Enterprise and is priced separately, unless both services are covered by a consolidated enterprise agreement.
THE UNIT OF USAGE
A governed decision is one unique, authenticated decideAction or decideToolCall request accepted by a workspace for processing. Policy may allow, refuse, or ask a person first. When required evidence is incomplete, the request is Not decided.
Approvals are included. A safer policy should not create a higher transaction charge.
PREDICTABLE GROWTH
We notify you as usage approaches the included allowance. Ordinary growth does not trigger an immediate production cutoff.
We review sustained overage with you and agree the appropriate capacity or plan before changing the commercial entitlement.
INDEPENDENT CONTROL
Documents, prompts, generated content, and raw action payloads remain in the supervised application. Agent Control Room receives the declared action, canonical resource reference, payload fingerprint, and the facts required by policy.
PAID VALIDATION PILOT
A bounded six-to-eight-week pilot can validate the integration, policy model, and failure behavior before live deployment.
Scope, fee, and any credit toward an annual plan are agreed before the pilot starts.
Start with the agents, action types, and environments that matter most.
Test policy, evidence, approvals, failure modes, and signed grants before production.
Agree the plan, capacity, retention, and deployment terms for live governed actions.
GOVERNED AUTONOMY
Start in a sandbox, validate the integration, and move to production with a policy and evidence model your organisation can explain.