// AI NEWS · AI VIDEO

HeyGen Pushes Past Avatars Into Full Shows: Video Podcast Turns a Document Into a Two-Host Video Episode

HeyGen launched Video Podcast, an app that writes the dialogue, casts two AI hosts who share a studio and react to each other, cuts between cameras, and renders a publish-ready 16:9 episode from a topic, URL, PDF, or audio track — part of its 2026 move beyond talking-head avatars into broader creative tooling.

2026-08-18 · by Moe Ameen

What happened

HeyGen — the identity-first AI avatar company that said it crossed $200 million in ARR in June 2026 — launched Video Podcast in 2026, its clearest step beyond single talking-head clips into full, produced shows. You give it a topic, a URL, a PDF, or an existing audio track, and it writes a conversational script, casts two AI hosts with distinct voices and timing, and renders a 16:9 HD episode. The hosts share one studio scene and react to each other rather than appearing as two separately generated clips stitched together, and the app cuts between wide establishing shots and per-host close-ups with B-roll layered over the dialogue, plus automatic captions and chapters.

The workflow keeps a human in the director's chair. You can edit any line and the voice regenerates automatically, preview the full episode before spending credits on the final render, adjust tone and delivery, and match the backgrounds and colors to a brand. HeyGen also added a co-host-by-email invite: a colleague or guest can join with their own avatar without needing a HeyGen account. Finished episodes can be localized into 175+ languages with preserved lip-sync, and export as standard video files for YouTube, a website, or social platforms.

Video Podcast is one piece of a wider 2026 expansion. Over the same stretch HeyGen shipped its open-source HTML-to-MP4 engine HyperFrames, its Video Agent for prompt-to-video, the more realistic Avatar V, and Instant Highlights for turning long recordings into short clips — a company that started with one talking avatar reshaping itself into a broader AI creative suite. Reports placed the Video Podcast rollout in the summer of 2026; treat the exact launch day as approximate and confirm on HeyGen's site.

Why it matters for creators

  • The two-hosts-in-one-scene approach fixes the biggest tell of AI podcasts — separately rendered avatars talking past each other. A shared studio with real camera cuts reads as a recorded show, not a slideshow.
  • Document-to-episode collapses production: a report, a blog post, or a PDF becomes a produced video conversation in minutes, with no studio, second person, or editor.
  • It is a scope expansion, not just a feature. HeyGen is moving from "make one avatar clip" toward "make a whole show," which changes who competes with it — podcast and repurposing tools, not just avatar generators.
  • Localization into 175+ languages means one episode can serve global audiences without re-recording, a genuine advantage for anyone publishing across regions.
  • The last mile is still unsolved: Video Podcast renders one long episode and stops. Clipping it into short-form, writing the promo posts, and scheduling across platforms are all left to the creator.

How to act on this with Kompozy

Here is the practical problem the launch creates. A 20-minute two-host episode is a beautiful asset and, on its own, a dead end — the discovery happens in the vertical clips and the promo posts, and Video Podcast makes none of those. The moment the render finishes you are back to a manual scramble: pull the best 40 seconds, cut it to 9:16, caption it, write something for LinkedIn, remember to actually post it. That gap between "episode exists" and "episode reaches people" is where Kompozy lives.

Drop a HeyGen episode into Kompozy and it runs the whole back half on autopilot. Clipped Shorts pulls the strongest moments and outputs captioned, auto-reframed verticals for TikTok, Reels, and Shorts; Quote Graphics turn the best lines into shareable cards; a Blog Article becomes the show-notes page and an Email Newsletter becomes the episode digest — every piece written in one voice through your Persona Brief. Then Kompozy schedules and publishes the full set across Instagram, TikTok, YouTube, LinkedIn, X, Facebook, Pinterest, and Threads from a single queue. Point Kompozy's autopilot at the source you feed HeyGen and the cadence runs itself week after week: new episode in, a full week of on-brand posts out. HeyGen just made the episode trivial to produce; Kompozy makes it trivial to turn that episode into a content week that actually gets seen.

Quick takeaways

  • Video Podcast turns a topic, URL, PDF, or audio track into a two-host 16:9 video episode with a shared studio scene and multi-camera cuts.
  • Edit any line (the voice regenerates), preview before render, and invite a co-host by email who joins with their own avatar.
  • Episodes localize into 175+ languages with lip-sync and export for YouTube, web, or social.
  • It renders the episode but does not clip, repurpose, or schedule it — Kompozy is the engine that turns one episode into a published content week.

Frequently asked questions

What is HeyGen Video Podcast?

It is a HeyGen app that generates a two-host video podcast from a topic, URL, PDF, or audio track. It writes the dialogue, casts two AI hosts who share a studio scene, cuts between cameras with B-roll, adds captions and chapters, and renders a publish-ready 16:9 HD episode you can edit line by line before rendering.

How many hosts does HeyGen Video Podcast use, and can I add a guest?

It uses two AI hosts who share one studio scene and react to each other. You can also invite a co-host by email, who joins with their own avatar without needing a HeyGen account.

Can HeyGen Video Podcast clip the episode and publish it to social media?

No. It exports a full episode you upload yourself, but it has no clipping engine or multi-platform scheduler. To turn one episode into captioned vertical shorts, promo posts, a show-notes blog, and a newsletter and publish them on a schedule, use a content engine like Kompozy, which does that across nine platforms from one queue.

Related news

← All AI news · Get started →