Synthesia: AI training product expands into interactive conversation practice

Synthesia launched Roleplay Sessions on July 22, 2026. The new product turns the company’s AI avatar technology from passive video generation into interactive conversation practice — employees can rehearse high-stakes workplace scenarios with an AI avatar that responds, challenges them, and scores their performance against predetermined rubrics.

The initial scenario types cover sales pitches, customer complaint handling, and difficult conversations such as performance reviews. The avatar adapts within the conversation rather than following a fixed script, and the platform generates analytics that companies can use to measure training outcomes at scale. CEO Victor Riparbelli describes the underlying rationale: “Video performs way better than text… but for most things, we learn the best by actually practicing something rather than just reading it.”

For product managers working on L&D or HR products, the launch signals a pattern worth tracking. Synthesia is building performance analytics on top of its existing avatar infrastructure — the moat shifts from “can you make a convincing avatar” to “can your rubric system accurately assess whether someone is improving.” This is a different kind of retention and stickiness than video courses provide, since scoring data creates organizational memory about individual skill development over time.

The expansion roadmap includes job interviews and candidate screening, which Riparbelli notes will become viable as inference costs continue to fall. For enterprise product teams, the case here is direct: Synthesia started with content generation, identified a higher-value adjacent use case, and is extending the platform while the same underlying AI infrastructure handles most of the execution.