HeyGen is an AI video platform built around a digital presenter: a person on screen who speaks a script with matching lip movement. You type the script, pick a presenter, and the video renders, often within a minute according to HeyGen. It is used for training, marketing, product demos, e-learning and sales outreach. Everything about HeyGen on this page comes from its own site, checked on 29 October 2026.
What HeyGen does
- Ready-made presenters. A library of over 700 stock presenters, cleared for commercial use.
- Your own likeness. About 30 seconds of video of a person talking to camera makes a personal avatar, usually ready in 10 to 20 minutes. One front-facing photo makes a photo avatar.
- Slides and screen recordings. A slide deck or PDF becomes a video with the presenter in the corner, and a presenter can be placed over your own footage or a screen recording.
- Translation. An existing video is translated into any of 175+ languages and dialects, with the speaker's voice cloned and the lips matched, up to ten languages at a time.
- Live conversation. LiveAvatar, previously called Interactive Avatar, is a presenter that talks with people in real time, used for customer support, tutoring and training agents.
- More than one presenter. Up to three avatars can share a scene, or a conversation can alternate between scenes with one speaker each.
What HeyGen costs
Monthly prices on HeyGen's site, in US dollars:
- Free: videos up to one minute.
- Creator, $29: videos up to 30 minutes, exported in 1080p.
- Pro, $49: videos up to 30 minutes, exported in 4K.
- Business, $149: videos up to 60 minutes, five custom avatars, $20 a month for each extra seat.
- Enterprise: priced on request, with no maximum video length.
Where HeyGen stops
Every HeyGen video is built around a person speaking. The presenter can gesture and change expression, be framed wide, medium or close, and be directed on gaze, posture and energy with text prompts. HeyGen's pages don't describe a person walking through a workplace, handling a tool, or two people working on the same job. A presenter placed over footage of a factory floor sits in front of that footage as a layer.
A lot of training fits inside that: a safety briefing, a policy explainer, a welcome from the CEO, or the same message in six languages.
When the video has to show the work
Some training videos need to show the job being done: a mechanic working on a bus, hands on a piece of equipment, a person moving through a site. That kind of video is what I make through AI video production, and equipment maintenance training videos covers how a machine is built from photos.
When the training needs more than one character in a scene, with the same cast across a series, that's covered on quality awareness training videos.
Each shot starts as a still image made with AI image models and checked against photos of the real equipment and place. Seedance 2.5 then animates the still into a clip of five to ten seconds, the narration comes from an ElevenLabs V4 voice, and everything is put together in a conventional edit. The same presenter and location can carry through a whole series.
It takes longer than a presenter video. Each clip takes five to ten minutes to generate and often needs another go, and most of the time goes into the last details. I wrote about that process in making AI training video with Seedance 2.5 and ElevenLabs V4.
Tell me about the video.
Send what the video needs to show and who will watch it, and I'll come back with a quote against scope.
Ask for a quote →