Synthesia‘s Roleplay Sessions launch marks a sharp pivot for the British AI startup: from a company that makes training videos to one that scores your performance in them. The product, released this week, drops employees into simulated high-stakes conversations with an AI avatar, then grades them on a rubric.
The use cases include sales pitches, performance reviews, and customer complaints. The avatar talks back, pushes back, and generates data for managers. Neat demo fodder, yes, but also a genuine attempt to fix what Synthesia openly admits is a problem with its own core product.
Why Synthesia Is Betting on Roleplay Sessions
Synthesia cited a meta-analysis of learning research in arguing that conventional corporate training, including video training, stops short of changing behaviour. Watching a demonstration informs. Actually practising the thing, with feedback, is what shifts performance. ‘Video performs way better than text or sending out a document,’ CEO and co-founder Victor Riparbelli told TechCrunch. ‘But for most things, we learn the best by actually practising something rather than just reading it.’
The Roleplay Sessions launch is the first release under a broader ‘Sessions’ platform Synthesia plans to expand into job interviews and candidate screening. For now, though, the focus is enterprise: the company already has scaled commercial rollouts with ‘one of the top three companies by market cap in Europe, one of the top five Fortune 100, and one of the biggest recruitment companies in the world,’ according to a company spokesperson. Synthesia’s LinkedIn page lists over 90% of the Fortune 100 among its customer base.
The two most popular programmes so far involve sales training and leadership development, particularly soft skills: how to sell a product, how to have a difficult conversation. Customers can build their own rubrics on the platform or bring in Synthesia’s consultants to construct something bespoke from their existing training documents.
From Avatar Company to Performance-Management Platform
The strategic logic is reasonably clear. Synthesia’s avatar and voice technology are proprietary, but the underlying reasoning intelligence is OpenAI’s. That’s a thin moat. Layering on rubrics, performance analytics, and a manager dashboard repositions Synthesia as a performance-management platform with a video front end. It’s a more defensible place to sit.
Riparbelli is specific about what that means for enterprise buyers. ‘If you have your entire sales team practise with an AI role player, you can get very granular data on how your sales force is doing overall,’ he said. ‘From a managerial executive layer, we’re seeing huge interest in mapping out the talent in their company in a way that’s been really difficult to do before.’
The ambition extends beyond enterprise over the next few months. Riparbelli said Roleplay would expand to small businesses, prosumers, and potentially educational institutions as inference costs continue to fall.
All of which arrives at an interesting moment for Synthesia’s finances. The company raised a $200 million Series E in January 2026, valuing it at $4 billion, nearly double the $2.1 billion valuation it carried a year prior, according to a Synthesia Series E announcement. The round was backed by GV (Google Ventures) and nVentures (NVIDIA), per Observer. That valuation jump came against a broader surge in European AI investment: the continent’s AI companies raised $17.5 billion in venture capital in the year before Synthesia’s raise, up 75% year-over-year, according to Crunchbase data cited by Observer.
Synthesia was founded in 2017 by Riparbelli, Steffen Tjerrild, Matthias Niessner, and Lourdes Agapito, drawing on AI research from Stanford, Cambridge, UCL, and TUM. Its platform currently supports avatar video creation in over 160 languages, with the company claiming users save up to 90% of the time and cost associated with traditional video production.
The bigger question hanging over Roleplay Sessions, and over enterprise AI broadly, is whether ‘measurable outcomes’ is a genuine product category or a marketing answer to a CFO’s question. Synthesia has at least structured this one so the data comes back to the buyer, not just the vendor. Enterprises paying for AI training programmes will now have rubric scores to scrutinise. Whether the scores mean what Synthesia says they mean is the next argument to watch.
