The AI Agent action has been updated with three improvements - a built-in Call Transcript tool, a structured feedback flow in Execution Logs, and a round of bug fixes and stability improvements.

Built-In Call Transcript Tool

The AI Agent action now includes a built-in Call Transcript tool that gives the agent access to a contact's full call history. The tool works internally with no configuration required - simply reference call transcripts in your agent prompt and the agent retrieves and uses them automatically.

For example: "Use this contact's call transcripts to summarise their concerns" - and the agent handles the rest.

Combined with the previously released conversation history tool, the agent now has access to both past chat conversations and call transcripts - a more complete view of every contact interaction before it responds.

Feedback Flow in Execution Logs

You can now submit structured feedback on every AI Agent run directly from the Execution Logs. Flag what went wrong, note what worked well, and share specifics for each run. All feedback is used to improve the AI Agent's accuracy, evaluation quality, and goal alignment over time.

This closes the gap between what the agent does and what you expect - enabling faster iteration and fewer deviations.

Bug Fixes and Stability Improvements

A number of bugs across the AI Agent action have been resolved to improve reliability and overall experience.

Why It Matters

An AI Agent that can only read chat history and contact fields is working with an incomplete picture. Call transcripts capture voice interactions that don't show up anywhere else - the concerns raised, the objections given, the context shared in a conversation. With access to that, the agent's decisions, responses, and actions are grounded in a much more complete picture of each contact relationship.

If you have any questions about AI Agent actions inside ESC Hub, the support team is here every day. Drop a message inside Escapepreneur Central.