Governed agents / Templates
Ten agents for approval-heavy work, each written as a short contract: what it can do, what it cannot, what needs a person, and what it records.
These are the shapes we would build and that you can build in Studio today. Reading them in this form is useful even before you build anything: it is the conversation to have with the process owner and the control owner.
None of these ten ships as a ready-made template today. Five have a starting point: a related template or Worker exists and the approval gate is yours to add. Five are designed only. We say which, on each card.
How to read the status
Each one lists what it can do, what it cannot do, what needs a person to approve, and what it records.
Prepares purchases for approval. It reads, compares and drafts, and it never approves or pays for anything itself. This is the worked example used across the platform pages.
Starts at: L2 Approve: a person approves each purchase it prepares.
No ready-made template ships yet. The shape is the one the platform pages describe.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Reads a support ticket and the order history, and recommends whether a refund is warranted. A person or a payments system issues it.
Starts at: L1 Assist, moving to L2 Approve for refunds once the recommendations are reliable.
The agent wizard includes a customer-support sample, and workflows have a human-input step you can use as the approval gate. The refund policy and gate are yours to add.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Prepares periodic access reviews. It finds access that looks stale or too broad and puts a clear pack in front of each reviewer.
Starts at: L2 Approve.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Prepares a change request for its approver: what changes, what could break, and whether the evidence is complete.
Starts at: L1 Assist.
Studio has Worker templates for code review and DevOps. They produce assessments and recommendations only, with no approval flow, so the gate is yours to add.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Compiles the evidence for a release decision and recommends go or no-go. A named person decides.
Starts at: L2 Approve.
A release-manager Worker template exists. It is advisory: it recommends, and nothing in it publishes or approves.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Helps the on-call team in the first minutes: correlates alerts, drafts a timeline and proposes first steps. People take the actions that change things.
Starts at: L1 Assist, with containment staying at L2 Approve.
Security analyst and production-support Worker templates exist, as does a paging integration. They triage and recommend. The approvals and the timeline record are yours to add.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Checks a new supplier file for completeness and summarises the risk answers for the person who decides.
Starts at: L2 Approve.
No template exists. The nearest shipped one is a contract-review workflow with named sign-offs.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Assembles a know-your-customer case for a trained reviewer. It organises evidence and never decides.
Starts at: L1 Assist. Decisions stay with people.
No template exists. A KYC onboarding listing on this site is a sample listing, not a shipped template.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Collects what a data protection impact assessment needs and drafts the first version for the data protection officer.
Starts at: L1 Assist.
No template exists. DPIA appears in the platform only as a control and an attestation topic.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
Runs a policy attestation round: asks the named people, chases, and reports who has and has not attested.
Starts at: L2 Approve.
The certification feature records human attestations against controls. The chasing and the report are yours to assemble.
| Can | Cannot | Requires approval | Records |
|---|---|---|---|
|
|
|
|
The same five steps apply to any of them.
Agree Can, Cannot, Requires approval and Records with the process owner before you build anything.
Approval thresholds are the process owner’s call. Do not borrow a number from a template or a blog.
Begin with a person approving each action, and move up only on evidence.
Try every Cannot. If the agent can do one, the contract is not enforced yet.
After a trial, read the run record as an auditor would. If you cannot reconstruct a decision from it, fix the record before you widen the scope.
A template is a design, not a certificate. It does not make a process meet a regulation, and the right approvers, thresholds and retention periods depend on your organisation and your jurisdiction. Whether a template is available on your plan is a question for our team: <template availability by plan - founder to fill>.
Not as ready-made templates. Studio has the building blocks, such as agent nodes, a human-input approval step and branches, and some related templates and Workers exist. Five of the ten have a starting point and five are designed only. Each card says which.
Because trust in an agent is something you observe. Starting with a person approving each action gives you the evidence you need to move it up, and a clear point to move it down if the evidence turns.
The process owner. The templates say "above the threshold" and never give a number, because the right number depends on your organisation and risk appetite.
The templates say it cannot. A platform check that the approver differs from the requester is designed for and not built yet, so you enforce it when you assign each approval gate.
We make no such claim. They are designs for controls and records that can support the evidence an organisation needs. The exact posture depends on your use case, jurisdiction, deployment and configuration, and this is not legal advice.
Start with one entry point. Add intelligence, agents, workflows and infrastructure as you prove value.
One platform for the agents, models and workflows your team ships. Free to start, no card required.