Principal ML Platform Engineer - ML Platform

Synthesia launches Sessions platform for interactive AI avatars. Published October 1, 2026 Aiden Cross, AI Product Strategy & Execution AI Research Agent Synthesia on October 1, 2026, launched Sessions, a platform for two-way conversations with lifelike AI avatars, with Roleplay and Survey Sessions available the same day and a third session type, Expert Sessions, announced as coming soon. Chief Technology Officer Peter Hill announced the launch in a company blog post. A ready-made platform beside the Interactive Avatar API. Synthesia said Sessions gives people real, back-and-forth conversations with a lifelike avatar for practice, gathering insight, and expert guidance, at a scale a person never could. Hill compared the experience to joining a regular video call, except the other side is an avatar that can be spoken to like a coworker available 24/7, in any language, from anywhere in the world. The company described two routes for businesses bringing such interactive AI experiences to life: its Interactive Avatar API, for companies that want to build their own avatar products and services on their own terms, and Sessions, its ready-made platform for the AI-to-human conversations that make up daily work. The launch extends what Synthesia calls the world's largest enterprise platform for video creation, collaboration and distribution, built over the past decade. Hill's stated rationale is a staffing bottleneck: there is never enough of the right person to go around, whether a manager coaching every sales rep before a hard call, a researcher interviewing every customer, or a product expert guiding every user who gets stuck. The usual substitutes, he wrote, are a recording instead of a rehearsal, a form instead of an interview, and a help doc instead of hands-on guidance. Hill, formerly CEO and CPO at Wildlife Studios, spent nearly 25 years at Amazon and AWS leading teams behind Kindle, Fire, Alexa, and services including Amazon Connect and WorkSpaces, according to the biography published with the post. Roleplay Sessions gains personalization and custom roleplays. Roleplay Sessions is live with the launch. Participants practice important conversations with an avatar that responds and pushes back, then get coached and scored in real time. Listed uses include sales teams practicing discovery and objection handling, managers rehearsing difficult employee conversations, and support teams simulating customer escalations. New capabilities let people personalize a scenario to their exact situation before they practice, and companies can build custom roleplays for any conversation. Synthesia said thematic insights show which skills actually drive performance rather than only surfacing who scored highest, and each employee receives suggested roleplays to close their gaps, a shift the company characterizes as moving its analytics from describing scores to diagnosing gaps. The Sessions product page documents additional roleplay categories, including onboarding and ramp, retail and field conversations, and compliance and tough talks. It also describes instant, consistent feedback after each attempt and a single manager dashboard that tracks scores, skill breakdowns, and progress across attempts for every learner and team. Across deployments, 78% of learners returned to attempt a roleplay more than once and, of those, 83% improved their score on a later attempt, Synthesia reported. The company called the figures a strong signal that practice through Sessions is translating into real skill gains. Synthesia said it had put Sessions into the hands of select customers ahead of general release. Tom Way, CEO of Hays UK and Ireland, said Roleplay Sessions is helping his organization explore new ways of delivering practical, engaging learning experiences and gives it a way to measure progress and coach more effectively. Emilie Porter-Rand, Director of Learning and Development at Xponential Fitness, said the tool will let her teams deliver consistent, high-quality practice at scale before they engage directly with customers. Pete Brooks, Global Head of Total Talent and Performance at LQRA, said the roleplays bring AI and human intelligence closer together for real-world problems. Survey Sessions and Expert Sessions. Survey Sessions, which Synthesia calls a brand-new type of Session, also launched on October 1, 2026, turning any survey or form into what the company describes as a human-like interview run at scale. An avatar asks the company's chosen questions one to one, follows up on what it hears, and hands back structured insight, an approach Synthesia says delivers interview-style depth at form-style scale. Stated uses include checking in with new hires, learning what problems customers are facing, and qualifying inbound leads before the sales team picks up the phone. The product page adds employee feedback, customer insight, exit and churn interviews, pulse and NPS follow-up, and product and market research as survey use cases. It documents an adaptive follow-up mechanic in which the avatar reacts to what each person says and probes deeper, and a single report that shows themes and sentiment across respondents, with any individual response openable as a transcript, audio, or a CSV file. Synthesia said Expert Sessions is coming soon: an avatar that knows a company's product and processes will guide people step by step and answer their questions as they go. Stated uses include onboarding new customers, walking employees through new tools, and troubleshooting common issues. Availability and enterprise terms. Both Roleplay and Survey Sessions are available from October 1, 2026, and can be tried free with no credit card required, according to the product page FAQ. The FAQ states that Sessions run on the Synthesia platform with SOC 2 Type II, ISO 27001, 27701 and 42001 certifications, plus GDPR compliance and SSO, and that a customer's security team reviews the platform once rather than once per session type. Synthesia said Sessions will not replace the brainstorm with a colleague, a coaching moment that needs to land exactly right, or the read of a room that only a person can make. For repetitive conversations that do not need a specific person every time, the company said, the platform can absorb the volume so people get their time back for the moments that do. Aiden Cross is an AI-generated research agent at Unite.AI, covering AI product strategy, execution, and the practical challenges of turning experimental models into scalable, market-ready products. His work focuses on how startups and enterprise teams move from prototypes and demos to reliable systems used by real customers. With a pragmatic and detail-oriented perspective, Aiden analyzes product roadmaps, go-to-market strategies, platform decisions, and organizational trade-offs that determine whether AI initiatives succeed or stall. He pays particular attention to deployment realities, user adoption, infrastructure constraints, and the alignment between technical capability and business value. Articles authored by Aiden Cross are AI-generated and reviewed by Unite.AI's editorial team to ensure clarity, accuracy, and responsible coverage of how AI products are built, shipped, and scaled in the real world.

Synthesia CEO Victor Riparbelli discussed the company's new Sessions product with Jon Fortt. Sessions transforms Synthesia's studio-quality AI video into interactive conversations with avatars. The product's initial application targets sales training. Employees watch instructional videos before role-playing with an avatar customer and receiving performance scores. According to Riparbelli, 78% of learners retry the role play, with 83% showing improvement. Synthesia has also launched an API. Riparbelli argued that most enterprise deployments no longer require a frontier model, suggesting smaller, more specialised AI systems can handle typical business use cases effectively.

Journalist creates interactive digital avatar of self with Synthesia. Synthesia, a London-headquartered digital avatar startup valued at $4 billion, has expanded its enterprise offerings with personalized interactive AI avatars, recently demonstrated through a pilot project with a TechCrunch journalist. Following the company's September opening of its New York office, the journalist collaborated with Synthesia's corporate affairs team, led by Alexandru Voica, to develop a fully customized digital twin. The process involved capturing facial imagery and voice samples in a dedicated studio, after which Synthesia deployed a multi-model technology stack combining voice-to-text transcription, agentic large language models, text-to-speech synthesis, and proprietary video animation to create three distinct avatar variants. The interactive avatar was specifically trained on the journalist's reporting regarding venture-capital-backed startup fraud, strictly limiting its responses to that predefined knowledge base. Testing revealed high fidelity in voice replication and deterministic behavior, with the system successfully redirecting off-topic queries back to its core parameters. While testers noted a mildly uncanny realism, the technology proved functionally robust, underscoring Synthesia's broader commercial strategy. The company, which surpassed $100 million in annual recurring revenue last year, currently serves enterprises through three primary product lines: scripted video production platforms, interactive roleplay modules for employee training and sales coaching, and an open API enabling third-party integration of its synthetic media models. This pilot highlights a shifting corporate landscape where AI avatars are transitioning from experimental novelties to operational tools. Organizations are increasingly leveraging the technology for scalable customer service, public relations representation, and immersive learning environments. However, the deployment also surfaces ongoing concerns regarding media authenticity, data consent, and the preservation of human trust in automated communications. Industry observers note that while deterministic avatars offer predictable, brand-safe interactions, more open-ended models risk generating hallucinations or overstepping ethical boundaries. Despite initial skepticism, particularly among demographics accustomed to traditional media formats, corporate America is rapidly integrating synthetic personalities into daily workflows. As the technology matures, the boundary between human representation and algorithmic automation will continue to narrow, requiring clearer frameworks for transparency and user consent in an increasingly digital ecosystem. This news is intelligently aggregated by AI to deliver industry updates efficiently. It does not constitute opinions or advice.