Skip to content
MirrorFirm

What is a useful pilot when an agent can update one system but another action fails?

A delivery access pilot can test whether an agent records partial completion and selects a defined recovery path when a downstream update fails.

A logistics operations team scopes a pilot around customer requests to change a site's unloading entrance. The agent can update site record SITE-2187 and request a revised driver instruction for job JOB-0412. The site update succeeds, but the driver instruction service refuses the revision because the job has entered a restricted dispatch stage.

The pilot defines what the agent should do next: retain the valid site change for later jobs, mark the current job as unresolved and create a dispatcher review request. Reverting the site record would discard a useful future instruction without fixing the current delivery.

The buyer asks for the initial state, the downstream refusal and the permitted recovery actions in the sample. Acceptance depends on the agent representing partial completion accurately and directing the remaining work to the responsible role.

This scope makes a practical limitation explicit: the agent cannot always make every dependent system agree immediately. The pilot tests whether it handles that limitation according to a defined business rule, including communicating the unresolved consequence.

Agree the business treatment of partial completion before using failure cases to judge agent quality.

Audience:
Agent developers
Content type:
Scenario
Buyer question:
Scope and getting started

Content release: r-a220eb6e099b243b

More on Customer change handling