Release notes

What's changed in TeachPilot, written for the people who use it.

A clearer studio, and conversations that start the way you authored them

This update closes a set of gaps between what you author and what learners actually experience.

New

Release notes

This page. Each update to TeachPilot now gets a short summary of what changed for coaches and learners, so you are not left inferring it from the product.

New

The studio shows what the engine will actually do

Until now the scenario panel showed what you authored and nothing more. It now also shows what the engine derives from it: which agenda items can hold a conversation open, which the persona will steer toward, how the conversation opens, and which personas and rubrics resolved to a published version. Behaviour that used to be visible only by running the scenario is now visible while you build it.

Fixed

A persona's background now shapes how they behave

The background you write for a persona was being saved and then ignored — it reached no part of the conversation. It now informs how the persona responds, so what you write in that field does what the field implies.

Fixed

Voice conversations that open with the persona now speak

A conversation set to open with the persona left learners in voice mode listening to silence: the opening text existed but was never spoken. The persona now speaks its opening aloud, so a voice scenario no longer looks broken at the moment it starts.

Improved

Generated scenarios can open with the persona speaking

Scenario generation could only produce conversations that wait for the learner to speak first. It can now author a persona-led opening, so a generated scenario matches how the situation would really begin.

Fixed

Sessions move forward reliably

Moving between stages is now handled one step at a time, so a slow or repeated request can no longer skip a stage, apply twice, or roll a learner's progress backward.

Improved

Conversations start faster

The first reply in a conversation with a coverage agenda was paying a full cold start. That preparation now matches what the conversation actually needs, so learners wait less for the opening exchange.