Plan allowances and capabilities are shown below. Monetary pricing is not yet published. Join the waitlist
Plan fit
Find the operating model that fits.
Adjust your expected usage and operating requirements.
Compare plans
Choose your operating model.
Governance stays consistent. Capacity, collaboration, and organizational scope grow with you.
The engine−
Capability
Developer
Pro
Scale
Observe ModeEvaluate policy and record what Aethis would have done, while execution continues unchanged.
Allow and block policiesEvery governed action is evaluated against a versioned policy before it runs.
Conditions on every ruleEnvironment, resource, argument value, threshold, and agent identity.
Spend budgetsA spend limit you set on an Agent or workspace. A governance control, separate from the plan’s governed-action allowance.
Correlated Action RecordsDecision, policy, approval, execution, outcome, and cost, connected.
Volume−
Capability
Developer
Pro
Scale
Governed-action allowanceOne agent action that Aethis evaluates and records counts as one governed action, whether it is allowed, blocked, or held for approval.
10,000 / month
100,000 / month
1,000,000 / month
Higher volumeMove to the operating model sized for the amount of governed traffic and organization scope you need.
Move to Pro
Move to Scale
Organization-sized
Collaboration−
Capability
Developer
Pro
Scale
SeatsPeople who can access the workspace.
5
Unlimited
Unlimited
Shared review and approvalsA second person reviews and decides on a held action.
Not included
RolesRoles that match how your team reviews and administers governed actions.
Custom roles
Custom roles
Custom and organization roles
Reviewer eligibility and scopeLimit who can decide held actions instead of treating every administrator as interchangeable.
Not included
Memory and evidence−
Capability
Developer
Pro
Scale
Evidence retentionHow long Action Records stay available in Activity before they are deleted under the plan retention policy.
30 days
90 days
1 year or custom
Evidence exportMove Action Records into external review and audit workflows.
CSV and JSON
CSV and JSON
CSV, JSON, and streaming export
Organization control−
Capability
Developer
Pro
Scale
WorkspacesSeparate governance boundaries for teams, products, or environments.
2
2
Unlimited
Organization consoleCoordinate workspaces, members, approvals, and policy from one layer above.
Not included
Not included
Organization-wide governance baselinesSet shared governance requirements once while allowing workspaces to add stricter local controls.
Not included
Not included
Workspace-scoped administrationLet workspace owners manage their agents, routes, members, and local controls without organization-wide administration.
Not included
Not included
Delegated review scopesRoute approval responsibility to the people responsible for the workflow instead of a universal central reviewer.
Not included
Not included
Workspace evidence boundariesKeep one workspace from automatically gaining access to another workspace’s agents, evidence, or operating details.
Not included
Not included
Google and GitHub sign-inAvailable at every size.
Enterprise identitySAML and SCIM for organization sign-in and provisioning.
Not included
Not included
SAML and SCIM
Audited administrator accessAn organization administrator entering a workspace generates its own Action Record, visible to that workspace.
Not included
Not included
Clear counting
One action.
One evaluation.
A governed action is counted when Aethis evaluates and records it, whether it is allowed, blocked, or held for approval. MCP discovery, health checks, and pings are not counted, and an idempotent retry of the same action still counts once in total. Your plan allowance is governed actions per month, not spend; spend budgets are a separate control you set on an Agent or workspace.
Evaluated work
Actions
One model call through a governed routeOne evaluation, one decision, one record.
1
One tool call evaluated by AethisA tool call is an action like any other.
1
One model call that triggers 4 tool callsThe model decision is evaluated once and each tool call is evaluated on its own.
5
100 separate tool calls delivered in one batchEach one can be allowed or blocked independently, so each one is evaluated.
100
Approvals, retries, and other cases +
Evaluated work
Actions
One action moving through evaluation, approval, execution, and outcomeThose are stages of one record, not four actions.
1
An action blocked by policyThe evaluation ran and the record was written. Blocking is the product working.
1
An action held for approvalHeld and then approved is still one action.
1
An action observed but not enforced (Observe Mode)The evaluation and the Action Record will still happen.
1
A retry carrying the same idempotency keyThe key identifies the action, so the retry joins the original record.
1 total
A call blocked because a tool changed since discoveryDrift triggers a real policy evaluation and produces a real record.
1
MCP discovery, health checks, and pingsPlumbing is not an agent action.
0
A few things worth knowing.
What is included in every operating model?
The core governance path: versioned allow and block policy, conditions, spend budgets, approvals, and correlated Action Records, and Observe Mode. The operating models differ in scale, collaboration, retention, and organization control.
What is a governed action?
One agent action that Aethis evaluated and recorded. A blocked action and a held action each count as one. A model call that triggers four independently governed tool calls is five actions because every call can receive its own decision.
Why do blocked actions count toward volume?
Because the policy evaluation and Action Record still occurred. Counting only successful execution would make the usage model depend on policy outcome instead of the work Aethis performed.
Why do shared approvals begin at Pro?
Developer is aimed at proving a governed workflow with a small operating group. Pro is the point where review becomes collaborative and another person regularly owns approval decisions.
When do I need Scale?
When governance spans multiple workspaces or the organization becomes the administrative boundary: organization roles, enterprise identity, shared baselines, cross-workspace oversight, and longer or streaming evidence retention.
Why does evidence retention vary by tier?
Longer retention increases storage and review scope. The periods shown (30 days, 90 days, and 365 days or custom) are enforced per plan.
What does export give me?
Portable Action Record data for external review, investigation, reporting, or SIEM and warehouse workflows. The exact export destinations expand at Scale.
How does the plan-fit tool pick a tier?
It reads governed-action volume, seats, workspaces, shared approvals, retention, and enterprise identity, then recommends the smallest operating model that satisfies those requirements. It does not estimate a price.
Give agents room to act.
Keep control of what happens.
Bring your first workflow under policy. Build from there.