Autopilot usage pool for testing

Usage pool consumption for Autopilot for testers and Autopilot agent, determined by the interface invoked rather than the capability itself.

Autopilot for testers capabilities and Autopilot agent (Autopilot chat) draw from the same monthly usage pool, but not in the same way. What determines consumption is the interface you use, not the capability you invoke.

Two interfaces exist for the same set of capabilities: a dedicated action or button in Test Manager or Studio Desktop, and Autopilot agent, the conversational assistant available in both products. Invoking a capability through its own dedicated action never consumes the usage pool. Invoking the same capability through Autopilot agent always consumes the pool, because every agent turn draws from it regardless of what the turn accomplishes.

How usage pool consumption works

Autopilot agent consumes from the usage pool every time you use it: in free-form conversation, or when you ask it to run a capability on your behalf. When Autopilot agent executes a capability for you, that capability adds no extra consumption on top of the agent turn — the turn itself carries the entire cost.

• Invoked directly — Using a capability's own dedicated action or button in Test Manager or Studio Desktop never consumes the pool.
• Invoked via Autopilot agent — Asking Autopilot agent to perform the same action, in free-form conversation, consumes the pool for that turn. The capability running underneath adds nothing on top.

For general usage pool mechanics — plans, license tiers, and monthly allocations — refer to Licensing.

Capabilities available directly and via Autopilot agent

The following capabilities exist in Test Manager and Studio Desktop. Most are available both as a dedicated action and through Autopilot agent.

Onde	Recursos	Agent-only or direct-only
Test Manager	Search / conversational Q&A	—
Test Manager	Gerar casos de teste	—
Test Manager	Evaluate requirements	—
Test Manager	Create charts	Agent-only — no dedicated direct action exists
Test Manager	Importar Casos de Teste	—
Test Manager	Encontrar Testes Obsoletos	—
Test Manager	Generate test reports (Test Insights)	Direct-only — not exposed in Autopilot agent
StudioDesktop	Generate Expression	—
StudioDesktop	Generate C# Code (coded test case)	—
StudioDesktop	Fix validation errors	—
StudioDesktop	Generate (low-code) test cases	—
StudioDesktop	Search documentation	—

None of these capabilities ever consume the pool on their own, not even when invoked via the agent. Autopilot agent itself, the general-purpose conversational interface in Test Manager and Studio Desktop, is the only capability that draws from the pool, based on the token usage of each turn against the default model.

Observação:

Create charts has no dedicated direct action — it is available only through Autopilot agent, so using it always consumes the pool. Generate test reports (Test Insights) is the opposite case — it is not exposed in Autopilot agent at all, so it never consumes the pool regardless of how it's invoked.

How much of my pool does Autopilot agent use?

Since Autopilot agent is the only conversational interface that draws from the pool, most testers can use it extensively without materially depleting their monthly allocation. A single complex, multi-step interaction, spanning several agent turns across search, evaluation, and generation, typically consumes a small fraction of a Tester's monthly pool, and an even smaller share for App Test Developer tiers, which start from a larger allocation.

Example: a typical Autopilot agent interaction

The following example shows a typical Autopilot agent workflow for a tester:

• Create a chart of test cases per requirement.
• Find a requirement with no test cases through search.
• Validate that requirement's quality, then adapt it based on suggestions.
• Generate test cases for the requirement.
• Generate an automated low-code test case in Studio for one of those test cases.
• Fix a validation error that comes up during generation.
• Create a test set, assign the test cases, and run it.

The following table shows which of these steps consume from the usage pool.

#	Etapa	Capability invoked	Agent consumes?
1	Create a chart showing test cases per requirement	Create charts (Test Manager)	Sim
2	Find a requirement with no test cases	Search / conversational Q&A (Test Manager)	Sim
3	Validate the quality of the requirement	Evaluate requirements (Test Manager)	Sim
4	Adapt the requirement based on suggestions	— (manual edit, outside the agent)	Não
5	Generate test cases for the requirement	Generate test cases (Test Manager)	Sim
6	Generate the automated (low-code) test case in Studio	Generate (low-code) test cases (Studio Desktop)	Sim
7	Fix a validation error that comes up during generation	Fix validation errors (Studio Desktop)	Sim
8	Create a test set, assign the test cases, and run it	Create/run test set (Test Manager)	Sim

Consumption follows the interface, not the capability. Every step run through Autopilot agent consumes a turn, including steps 6 and 7, which run in Studio Desktop, while the one step done manually outside the agent (step 4) does not, even though it involves working with the same requirement. Even testers who run through workflows like this regularly are unlikely to approach their monthly pool limit, particularly at the higher App Test Developer allocations.