Synthesia, the AI video company that most enterprises know for building training content at scale, launched a fundamentally different product on July 22, 2026: Roleplay Sessions, an interactive AI coaching tool where employees have live conversations with AI avatars rather than just watching them.

The pivot is a meaningful signal about where enterprise learning is heading, and about what it now takes to prove that training worked.

What Synthesia Built

Roleplay Sessions puts employees in simulated workplace conversations with an AI avatar that adapts to what the employee says. Rather than passively watching a video, the employee is practicing real scenarios (handling a customer complaint, delivering a performance review, responding to a sales objection) with an AI that talks back, pushes back, and adjusts based on how the conversation unfolds.

At the end of each session, the system scores the employee against a predefined rubric and generates analytics the manager or L&D team can review. The goal is not just to expose employees to scenarios but to generate measurable data on whether the training changed their behavior.

The initial focus is on high-stakes, high-repetition conversations: sales pitches, difficult client calls, internal performance conversations, and customer service scenarios where slight differences in how an employee handles the situation have real business consequences. These are exactly the conversations that have historically been hardest to train at scale because they require personalized feedback, not passive content consumption.

Why This Matters Beyond the AI Video Market

Synthesia’s existing business, AI-generated training videos, is already embedded across 90% of the Fortune 100. The company raised a $200 million Series E in January 2026 at a $4 billion valuation, primarily on the strength of that content generation capability.

Roleplay Sessions is a different bet. The company’s explicit argument is that generating content is not the hard part anymore. The hard part is proving the content did something, specifically that it changed how an employee handles a real situation. Interactive AI coaching creates a feedback loop that passive video never could. Every roleplay generates performance data, which makes training measurable in ways that most L&D functions currently cannot achieve.

This matters for enterprise buyers because training ROI has always been difficult to demonstrate. Most corporate learning programs can show completion rates and quiz scores. Showing that an employee’s actual behavior improved in a real customer conversation is much harder. Roleplay Sessions is designed to capture that data directly.

What This Means for Business

Three angles are worth watching for enterprises running their own training programs.

The performance data angle is the real unlock. Completion certificates and video watch rates have never satisfied CFOs reviewing L&D budgets. A system that scores actual conversation quality against a rubric (how clearly the employee communicated value, how they handled a difficult objection, whether they reached the right outcome) gives training a number CFOs can evaluate against the cost of running the program. That changes the budget conversation for L&D teams.

Roleplay at scale replaces one of the bottlenecks in human coaching. Traditional roleplay training requires a trainer or manager to participate in the practice conversation, which caps how frequently employees can practice and how many can train simultaneously. AI avatars remove the capacity constraint. A sales team of 500 can all practice cold calling on the same day; a service team running 24 hours can practice shift by shift. The volume that would previously require dozens of trainers runs on infrastructure.

The skills landscape is shifting in exactly the way EDNA has been tracking. The ability to communicate clearly, navigate difficult conversations, and handle customer or stakeholder situations with judgment is becoming more, not less, valuable as AI takes on routine task execution. Roleplay Sessions is a bet that companies will invest in developing those judgment-heavy skills at the same time they are deploying AI to handle simpler work. That thesis is consistent with what enterprise buyers are telling us they need.

For Enterprise DNA clients building AI strategies, the Synthesia announcement is a useful data point: the companies with the most AI maturity are not just automating work, they are also investing in the human capabilities that AI cannot replicate and using AI to develop those capabilities faster than traditional training allows.

Roleplay Sessions is currently available for enterprise customers, with plans to expand to small businesses and educational institutions later this year.