One timeline that shows the conversation, the system change and the page the patient sees, with a single pass or fail.
Example scenario · Healthcare scheduling
A booking assistant that touches three systems and no one has tested end to end
Patients ask the assistant to move an appointment. It calls a scheduling API and the patient then looks at a web page. The team wants one test that follows the whole journey.
A small product team with a thin QA function and an assistant built from fast-moving generated code. This is a composite built from common patterns, not a customer story.
Where the team is
The product moves faster than anyone can read it.
The assistant, the scheduling system and the patient portal were each tested separately, when they were tested at all. A reply that says the appointment is moved is not the same as the appointment actually moving, or the portal showing it. Patient identifiers appear in conversations, so test data has to be handled with care.
What is hard to see
- Whether the appointment really changed in the scheduling system, not just in the assistant's reply.
- Whether the portal shows the new time to the patient.
- What the assistant does when the requested slot has just been taken.
- Whether real patient identifiers end up in test data.
The test, in plain steps
Test plan for rescheduling an appointment
Each step says what happens and what must be true. Anyone on the team can read it, challenge it and approve it, without opening the code behind the agent. An AI assistant can draft the plan over MCP, and a person reviews it.
1Set upCreate a test patient and an appointment in the staging scheduling system through its API.
2Mock toolcheck_insurance_eligibility is mocked, because it is a third-party service.
3Pass throughreschedule_appointment goes to the staging scheduling API.
4User says"Can you move my appointment to Thursday morning?"
5Assert replyThe assistant confirms the new time and does not offer medical advice.
6Read backFetch the appointment from the scheduling API. The new time is Thursday morning.
7Check pageThe patient portal shows the same appointment time.
What the team can show afterwards
Evidence anyone can read.
Traces turned into regression tests with patient identifiers masked first.
A readable test the clinic's operations lead can review, with no code to interpret.
A gate in CI so a change to any of the three pieces re-runs the whole journey.
More scenarios
Other places teams start.
E-commerce and retail
Small team, support agent
The agent answers order questions and can issue refunds. Nobody on the team has read every line the assistant wrote. Here is how they see what it does before customers do.
Financial services
Thinner QA, fintech copilot
A wealth copilot can check balances and prepare transfers. Risk and compliance still want evidence for each release, but fewer people are left to produce it by hand.
Developer tools and SaaS
AI-written code, agent swarm
An architect, coder, reviewer and tester agent hand work to each other. The delegation logic was generated, and the team wants to see and limit it without reading all of it.
Build this test for your own agent.
Request early access. Once your spot opens, you can have your first plain-step suite running against a staging endpoint in the first week of the 14-day trial.