For financial services
Show what your AI agent was allowed to do, on one real workflow.
When an AI agent acts in a regulated workflow, someone will ask who allowed it, whether a person approved it and what record it left. A paid proof of concept puts those questions to the test on one workflow you choose.
Book a free PoC scoping callDiscuss your workflow. The PoC is paid; scope and price are agreed before work starts. To change or cancel, reply to your calendar invitation.
The workflow problem
Problems people raise once an agent is working.
These come from recent conversations with practitioners and delivery partners. They are themes we heard, not survey results.
Access that outlives a role
Someone changes role, but the agent acting for them keeps the old permissions. Nobody notices until it uses them.
Approvals that turn into a queue
Sensitive actions wait for sign-off, and the wait pushes people to find a way round the approval.
Rules nobody can explain after a handover
The consultant or project team moves on, and no one can say why the agent is allowed to do what it does.
Exceptions with no owner
The agent does something unusual and it is unclear who answers for it, or what record shows what happened.
The control
What ACP can test on that workflow.
ACP checks actions that are routed through it. An agent action that bypasses ACP is outside these controls, so the first step is agreeing which actions will be routed.
Authority
Restrict the agent’s actions to the permission you agreed, including what happens when that permission expires or is revoked.
Human approval
Hold a sensitive action until a named person approves it, and record that they did.
Evidence
Keep a decision record for each allowed, denied and approved action, so a reviewer can inspect it later.
Failure
Test what the agent does when its permission has expired, has been revoked, or the control is unavailable.
Tests are limited to scenarios ACP supports today; the scoping call is where we check whether your workflow fits.
The evidence
Inspect it before you book.
Each page states its source, how to check it and its limits. They are product demonstrations and technical checks, not results from a customer.
The proof of concept
What you get, and what we need from you.
Scope, acceptance criteria and price are agreed in writing before any paid work starts.
What you receive
- A written test brief: the workflow, the actions in scope, and the success criteria we agree before paid work starts.
- A controlled demonstration of those tests on ACP, starting with synthetic data rather than your systems.
- A results note with the decision and signed-evidence references for each test, including any test that failed and why.
- A short review of what an integration in your environment would need, if you want to go further.
What you provide
- A workflow owner who can describe the AI agent’s actions and their consequences, without confidential data.
- Time for the scoping call and for reviewing the test brief and results.
- A decision at the end: stop, or discuss a longer-term service. There is no obligation to continue.
Installing ACP in your environment, cloud permissions and integration with your systems are separate stages, scoped, priced and accepted on their own. A demonstration on synthetic data does not show that a live workflow is protected.
Start with one workflow.
Bring a short description of the workflow and who owns it. Discovery happens on the call.
Book a free PoC scoping callDiscuss your workflow. The PoC is paid; scope and price are agreed before work starts. To change or cancel, reply to your calendar invitation.
Advising a client rather than buying? See the partner route.