Skip to main content

Inline Tool Calling

Understand tool-step status, approvals, results and generated outputs in Chat.

See how Hatz turns your request into a result. Tool steps show the work behind an answer, helping you understand which information was found and improve your next request.

Tools add capabilities to the AI model, such as web search or access to Outlook.

A tool lets Hatz search or act through a connected app. Open the relevant step in Chat to inspect the details shown:

  • What it’s searching

  • What tools it’s using

  • The responses from those tools

The new chat shows progress and results by step. Look for work in progress, an approval request, a completed result or a failure. Some details are shortened, and saved automation views can show different information.

Completed fictional Cedar calculation with the Ran Python step expanded to show code, output and exit status.

Inline tool calling can be helpful if you want to see how the AI handled a certain prompt, or how it used a tool, so that you can make changes in your prompts the next time.

Use what you learn

Check the tool’s search or action, the returned result and any generated file. If the result misses the information you need, refine the request with the source or scope you wanted. Before repeating a send or update, open the connected app and check whether the change completed. Tool troubleshooting explains recovery.

Follow a Python step while Chat works

While the calculation runs, the activity line shows the Python step and the square Stop control is available. Let the task finish, then expand the completed activity and compare the answer with the input: four, six and two hours total twelve.

Did this answer your question?