> ## Documentation Index
> Fetch the complete documentation index at: https://docs.stackshift.cloud/llms.txt
> Use this file to discover all available pages before exploring further.

# Evaluate and improve AI responses

> Compare AI behavior with examples and use outcomes to choose improvements.

<Note>
  This guide includes features awaiting release. If an option is not available in your workspace, contact StackShift Support.
</Note>

## What you can do

Compare AI behavior with examples and use outcomes to choose improvements. Where to find it: AI → Evaluations, Improvements and Procedure outcomes

## Before you start

* Permission to manage or review AI.
* Representative examples with the answer or behavior your team expects.

## Steps

<Steps>
  <Step>
    Create or open an evaluation and add realistic examples, including questions AI should hand to a person.
  </Step>

  <Step>
    Select the relevant configuration or procedure version and run the evaluation.
  </Step>

  <Step>
    Read individual results and their evidence. A combined score can hide one serious mistake.
  </Step>

  <Step>
    Use Improvements and Procedure outcomes to investigate repeated failure patterns; revise the knowledge, policy or procedure that caused them.
  </Step>

  <Step>
    Run the same examples again before broadening customer-reply scope. Review real outcomes afterward.
  </Step>
</Steps>

## Example

Include both a routine delivery question and a request for an unauthorized refund; expect an answer to one and escalation for the other.

## Things to know

* Evaluation results do not guarantee every future answer.
* An improvement suggestion needs review before changing a published procedure or customer policy.

## What happens next

<Check>
  You have a repeatable way to assess a change instead of assuming a new prompt is better.
</Check>

## If something goes wrong

<Warning>
  * Run cannot start: check provider availability, credits and inputs.
  * Good score but poor replies: include examples from the actual failing conversation and verify the evaluated version.
</Warning>

## Related guides

<CardGroup cols={2}>
  <Card title="Let AI answer customers" href="/stackshift-support/customer-ai">
    Enable customer-facing replies for a chosen scope, with rules for handing work to a person.
  </Card>

  <Card title="Support procedures and resolution" href="/stackshift-support/agentic-procedures">
    Create, test and publish support procedures with approved actions and human handover.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.