Synthesia, the London-based AI video company valued at $4 billion, has launched Sessions, a platform that turns its photorealistic avatars from one-way presenters into live conversation partners that can interview, coach and push back in real time.
Two modes are live today. Roleplay Sessions let users rehearse high-stakes conversations — a sales discovery call, a difficult piece of manager feedback — against an avatar that responds dynamically and, crucially, pushes back. Users receive real-time coaching and a score for each attempt. Survey Sessions replace static forms with a conversational interview: the avatar asks adaptive follow-up questions to dig into the "why" behind an answer, the kind of qualitative detail multiple-choice forms systematically miss.
A third tier, Expert Sessions, is on the way: avatars trained on a company's product documentation and internal processes, designed to walk users through technical troubleshooting step by step.
The company's own usage data suggests people treat these avatars more like colleagues than software: 78% of learners returned to retry a roleplay at least once, and 83% of those improved their scores on subsequent attempts — figures reported by Synthesia and not independently verified. CEO Victor Riparbelli tested the survey mode on his own workforce, having his digital double interview more than 700 employees with dynamic follow-up questions.
Under the hood, developers are not locked into Synthesia's stack. The Interactive Avatar API lets products embed real-time, lip-synced avatars driven by a company's own large language model, speech-to-text and text-to-speech systems — an acknowledgment that enterprises increasingly want the avatar as a skin, not the brain. The platform ships with SOC 2 Type II, ISO 27001, ISO 27701 and ISO 42001 compliance attestations and GDPR coverage.
The launch extends a busy run for the company, which counts more than 50,000 business customers — including, by its own count, 90% of the Fortune 100. The bet is that the next wave of enterprise video is not generated clips but always-on interactive ones: training that talks back, and surveys that feel like a conversation rather than a form.
Comments (0)
Log in to join the discussion
Log InNo comments yet