Agent identity
Know which agent and role initiated the request, which workflow invoked it and which human team remains accountable for the result.
AI infrastructure governance
Let agents move fast within explicit boundaries.
Acumy treats an AI agent as a distinct infrastructure actor, with its own identity, permissions, accountable team, budget, purpose and expiry. This preserves attribution even when infrastructure is created without a conventional human request.
Agent request · AGT-882Evaluating
Model evaluation environment
Identity verifiedScope matchedBudget attached12h expiry
Allowed with guardrails
A new class of maker
Machine-speed creation needs machine-speed governance
AI expands who or what can create infrastructure and compresses the time between intent and execution. Acumy makes that activity attributable, policy checked and finite without placing a manual ticket queue in front of every approved action.
Permission scope
Constrain approved workloads, templates, services, accounts, regions and access before the action reaches the cloud.
Dedicated budget
Give agent-created infrastructure an explicit financial boundary and keep its consumption separate from unauthorised or unexplained activity.
Time boundary
Apply a 12-hour default lease or a duration that matches your policy, with renewal treated as a new governance decision rather than an assumption.
Why agent identity matters
Automation should not become anonymous infrastructure
A service account or pipeline credential may reveal how a change was executed, but it does not always explain which agent made the decision, why the environment was needed or which team accepted the financial responsibility.
Acumy retains the complete chain of intent. The agent, initiating workflow, approved purpose, policy outcome, budget and expiry remain connected to the resulting Workspace.
Make the actor visible
Distinguish an agent request from a human request and retain the initiating workflow and accountable team.
Contain the possible action
Evaluate the requested workload, cloud scope, access, budget and duration before execution begins.
Preserve the decision record
Keep the original purpose and approval context available during operation, review, renewal and retirement.
AI agent
Automation
ACUMYIntent + policy
Same platform, distinct controls
Human and agent requests should not be indistinguishable
Acumy routes both through one operating model while retaining the controls each actor needs. An agent remains identifiable as an agent throughout the lifecycle, while a named team remains accountable for the environment it creates.
01
AttributeRecord the agent identity, initiating workflow, approved purpose and accountable team before execution.
02
ConstrainApply workload, service, access, financial and duration guardrails during policy preflight.
03
Observe and retireTrack the environment against its original boundaries and end temporary resources unless renewal is explicitly approved.
See Acumy in operation