Synthesia launched Roleplay Sessions on Wednesday, an interactive training product that lets employees rehearse sales pitches, performance reviews, and customer complaints with an AI avatar that pushes back and grades them against a rubric. The British startup is moving beyond the video-generation pitch that built the company and betting that the next enterprise buyer wants measurable behavior change, not just cheaper training clips. Early customers span one of the top three European companies by market cap, one of the top five Fortune 100, and one of the world's largest recruitment firms.
Roleplay is the first product under a broader Sessions platform. Synthesia plans to extend the format to job interviews and candidate screening, according to the company. The framing is a deliberate shift: instead of selling minutes of AI-generated video, Synthesia wants to sell proof that the training worked.
The company cites a meta-analysis of learning research to argue that most corporate training — including, implicitly, its own core video product — stops at informing and demonstrating rather than driving behavior change. Practice with feedback is what pushes someone from watching to doing.
Key facts
- 01Synthesia launched Roleplay Sessions on Wednesday, an interactive product where employees rehearse conversations with AI avatars that push back and score them against a rubric.
- 02Early customers include one of the top three European companies by market cap, one of the top five Fortune 100, and one of the world's largest recruitment firms.
- 03Reasoning inside Roleplay is powered by OpenAI, while the avatar and voice layer remain Synthesia's proprietary technology.
- 04The product is enterprise-only for now, with plans to expand to small businesses, prosumers, and educational institutions in the next few months as inference costs fall.
- 05Roleplay is the first release under a broader Sessions platform, which Synthesia plans to extend into job interviews and candidate screening.
That thesis is also a competitive hedge. Synthesia's avatar and voice technology are proprietary, but the reasoning layer under Roleplay comes from OpenAI. By stacking rubrics, performance data, and analytics on top of a foundation model it does not own, Synthesia is repositioning as a performance-management platform with a video front end rather than an AI avatar vendor competing on generation quality alone.
The two most popular use cases so far are sales training and leadership coaching — practicing product pitches or difficult conversations. Riparbelli said much of it is soft-skills work, the category that has historically resisted quantification inside large companies.
Customers can build their own programs on the platform or bring in Synthesia consultants to construct bespoke scenarios from existing training documents and internal context. The analytics layer turns each practice session into structured data on how individuals and teams perform.
Roleplay is enterprise-only at launch. Synthesia plans to open it to small businesses, prosumers, and potentially educational institutions in the next few months as inference costs continue to fall. That pricing trajectory matters: a rubric-graded conversation with an AI avatar is far more compute-intensive than serving a rendered training video, and margins depend on token costs continuing to drop.
The competitive picture is crowded. Every major model provider now ships some form of voice-mode conversational agent, and a wave of startups is chasing enterprise learning and development budgets with AI-driven coaching products. Synthesia's advantage is distribution — it already sells video training into large enterprises — and a proprietary avatar layer that most rivals have to license or build.
The risk is that the OpenAI dependency cuts both ways. If OpenAI ships a first-party enterprise coaching product on top of its own voice models, Synthesia's rubric-and-analytics wrapper becomes the primary defensible layer. That is exactly the bet Riparbelli is making by pushing into performance management rather than staying in content generation.
Synthesia's move maps to where enterprise AI budgets are heading. The demo-driven phase — buying AI because it looks impressive — is giving way to procurement teams asking whether the spend produced a measurable outcome. A sales team that practiced 200 objection-handling drills with scored feedback is a line item a CFO can defend; a library of AI-generated training videos with unclear completion rates is not. Synthesia is trying to be on the right side of that shift before the model providers get there first.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
