HeyGen
AIProduce AI avatar video at scale with HeyGen. Agents turn scripts into avatar-narrated videos, translate and localize existing footage, clone voices, generate lipsync, auto-clip long-form content, and manage avatars, assets, and webhook endpoints across 98 actions on the v3 API.
What This Integration Enables
HeyGen makes video cheap to produce, and that inverts the bottleneck. The scarce resource is no longer production capacity. It is judgment about what ships with your brand's face and voice on it. This connector gives FlowRunner agents the full v3 production surface: script-to-video with avatars and templates, video translation with voice cloning, lipsync, AI clipping, background removal, HyperFrames renders, asset management, and the webhook endpoints that report completions back. FlowRunner's job is to run that surface as a pipeline with the review moments built in. HeyGen's proofread sessions exist precisely so a person can correct a translation before the render, and the connector treats that as a first-class [human-in-the-loop](/concepts/human-in-the-loop) gate rather than an optional step. Agents move the files, poll the jobs, and assemble the batches. People approve what goes out wearing the company's spokesperson.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Localization with a proofread gate
A master product video lands in [Google Drive](/integrations/google-drive-api). The agent pulls it into file storage, uploads it with Upload Asset, and calls Create Video Translation for the target languages with voice cloning enabled. Instead of rendering straight through, it opens a Create Proofread Session per language, fetches the subtitle files with Get Proofread SRT URLs, and posts them to the regional reviewer in [Slack](/integrations/slack). The reviewer corrects the transcript, the agent submits it with Upload Proofread SRT, and Generate Video from Proofread produces the final localized video. Save Video to File Storage archives the render and the file goes back to Drive, reviewed before a single render credit was spent on a wrong sentence.
One script, a campaign's worth of video
The team maintains an approved template with variable slots for name, offer, and scene text. The agent inspects the schema with Get Template, fills variables from the campaign sheet, and submits up to 100 renders in one call with Create Video Batch. Get Video Batch and Get Bulk Video Statuses track the run, completed videos land in file storage, and the publish step pushes the hero cut to [YouTube](/integrations/youtube). The campaign gets personalized video at spreadsheet scale, from a template a human approved once.
Long-form recordings become a reviewed clip library
A webinar recording enters the pipeline. Create AI Clipping Job carves it into short highlight clips, Get AI Clipping Job reports the results, and the agent saves each clip with Save Video to File Storage. Rather than publishing everything the model liked, the agent delivers the clip set to the content owner with thumbnails and durations. The owner picks the clips that represent the brand, and only those move to the distribution workflow. The model proposes; a person curates.
Human-in-Loop Highlight
The gate in this connector sits exactly where HeyGen put it: between translation and render. Generate Video from Proofread spends render credits from a finite balance, and the output speaks in a cloned voice that belongs to a real person on your team. Neither of those should be triggered by an unreviewed machine translation. So the agent stops after Create Proofread Session and posts: "The German track for the launch video is ready. Review the subtitles here, original and editable SRT attached. Rendering uses the cloned voice of your spokesperson and consumes render credits; reply approve when the transcript is right." The reviewer edits, approves, and the pipeline resumes. What ships in your spokesperson's voice was read by a person first.
Agent Capabilities
98 actionsAvatar Videos
11- Create Avatar Video Creates an avatar video from a video avatar or photo avatar look. The core script-to-video action.
- Create Avatar Video and Wait Creates an avatar video and polls until generation completes, returning the finished video with its download URL. Used when the workflow needs the result inline.
- Create Video from Image Creates a talking video by animating an arbitrary photo instead of a pre-built avatar.
- Create Cinematic Avatar Video Generates a cinematic video from a natural-language prompt plus avatar look references.
- Get Video Retrieves a video's status, presigned URLs, captions, duration, and failure details. The polling step behind every render workflow.
- List Videos Lists account videos with cursor pagination.
- Delete Video Permanently deletes a video and its files. Used in retention cleanup, with review.
- Get Bulk Video Statuses Returns statuses for up to 100 videos in one request. Used to track large runs efficiently.
- Create Video Batch Submits up to 100 video payloads in one request and returns a batch ID. The scale lever for campaign production.
- Get Video Batch Returns aggregate batch status plus a page of member videos.
- Save Video to File Storage Downloads a completed video into FlowRunner file storage and returns a stable URL. The handoff step to publishing and archival workflows.
Video Agent
8- Create Video Agent Session Starts a session that produces a complete video from one natural-language prompt, handling scripting, avatar selection, and rendering.
- Get Video Agent Session Returns a session's status, progress, produced video ID, and recent messages.
- List Video Agent Sessions Lists sessions newest first with status and titles.
- List Video Agent Session Videos Returns all videos produced within a session.
- List Video Agent Styles Returns curated visual style templates for sessions.
- Get Video Agent Session Resource Returns a single resource produced within a session, with its URL and metadata.
- Send Video Agent Message Sends a follow-up message to a chat-mode session to answer questions or request revisions. The channel for iterating on a draft.
- Stop Video Agent Session Halts an active session at its next checkpoint.
Avatars
11- Create Photo Avatar Creates a photo avatar from a single photo.
- Create Digital Twin Avatar Creates a digital twin video avatar from recorded footage of a person.
- Create AI Avatar from Prompt Generates a fully synthetic avatar from a text prompt, optionally guided by references.
- Create Avatar Consent Initiates the consent flow required before a custom avatar of a real person can be used. The compliance step in twin creation.
- List Avatar Groups Lists avatar groups with pagination.
- Get Avatar Group Returns a group's previews, looks count, default voice, consent status, and training status.
- Delete Avatar Group Permanently deletes a private avatar group and all its looks. Gated in any cleanup workflow.
- List Avatar Looks Lists avatar looks with pagination.
- Get Avatar Look Returns a look's type, previews, supported rendering engines, and training status.
- Update Avatar Look Updates a look's display name.
- Delete Avatar Look Deletes a look and its backing resource.
Voices and Speech
7- List Voices Lists voices available for video generation and text-to-speech.
- Get Voice Returns a voice's language, gender, preview audio, and clone training status.
- Design Voice Finds up to 3 voices matching a natural-language description.
- Clone Voice Creates a voice clone from an audio sample. Consent-sensitive, and used only in workflows a person set up.
- Delete Voice Deletes a voice clone owned by the account.
- Generate Speech Synthesizes speech from text using HeyGen's text-to-speech engine. Used for voiceovers and audio prompts.
- Search Audio Library Semantically searches HeyGen's audio catalog for background tracks and effects.
Lipsync
8- Create Lipsync Replaces a video's audio track and re-animates the speaker's lips to match. Used to swap corrected audio without reshooting.
- Get Lipsync Returns a lipsync job's status and output URLs.
- List Lipsyncs Lists all lipsync jobs with statuses.
- Update Lipsync Updates a lipsync job's display title.
- Delete Lipsync Permanently deletes a lipsync job and its files.
- Create Lipsync Batch Submits up to 100 lipsync payloads as one batch.
- Get Lipsync Batch Returns aggregate batch status plus member items.
- Get Bulk Lipsync Statuses Returns statuses for up to 100 lipsync jobs in one request.
Templates
3- List Templates Lists the workspace's API-ready templates.
- Get Template Returns a template's variable schema and scenes. The inspection step before batch fills.
- Generate Video from Template Generates a video by replacing a template's variables with campaign values. The repeatable unit of on-brand production.
Translation and Proofreading
14- Create Video Translation Translates a video into target languages with voice cloning and lip-sync.
- Get Video Translation Returns a translation job's status, languages, output URLs, and captions.
- List Video Translations Lists translation jobs with statuses and outputs.
- Update Video Translation Updates a translation job's display title.
- Delete Video Translation Permanently deletes a translation and its files.
- List Translation Target Languages Returns all supported target language names.
- Get Bulk Video Translation Statuses Returns statuses for up to 100 translations in one request.
- Create Video Translation Batch Submits up to 100 translation payloads as one batch.
- Get Video Translation Batch Returns aggregate batch status plus member items.
- Create Proofread Session Extracts editable translated subtitles from a video before the final render, so a person can correct the translation. The foundation of this page's human gate.
- Get Proofread Session Returns a session's status, languages, and whether edited subtitles were submitted.
- Get Proofread SRT URLs Returns download URLs for the current and original SRT files. The step that puts the transcript in front of the reviewer.
- Upload Proofread SRT Replaces a session's subtitles with the reviewer's edited SRT file.
- Generate Video from Proofread Starts final rendering using the approved subtitles. Runs only after the reviewer signs off.
AI Clipping
4- Create AI Clipping Job Submits a long-form video and produces short highlight clips asynchronously.
- Get AI Clipping Job Returns the job's status, progress, and produced clips with download URLs.
- List AI Clipping Jobs Lists clipping jobs newest first.
- Delete AI Clipping Job Soft-deletes a clipping job and its clips.
Background Removal
4- Create Background Removal Submits a video for background removal.
- Get Background Removal Returns the job's status and, once complete, the foreground, mask, and background layers.
- List Background Removals Lists background removal jobs newest first.
- Delete Background Removal Soft-deletes a background removal job.
HyperFrames
4- Create HyperFrames Render Renders an HTML and JavaScript composition bundled as a zip into a video. Used for data-driven motion graphics.
- Get HyperFrames Render Returns a render's status, settings, and signed output URLs.
- List HyperFrames Renders Lists renders newest first with cursor pagination.
- Delete HyperFrames Render Soft-deletes a HyperFrames render.
Assets
10- Upload Asset Uploads a file from FlowRunner file storage or a URL to HeyGen and returns an asset ID for other actions.
- Get Asset Returns an asset's name, type, owner, and public URL.
- Delete Asset Permanently deletes an asset from the workspace.
- Search Stock Assets Semantically searches re-hosted public images and icons by description.
- Get Bulk Asset Statuses Returns statuses for up to 100 assets in one request.
- Create Direct Upload Begins a direct-to-S3 upload for files above the proxied upload limit.
- Complete Direct Upload Finalizes a direct-to-S3 upload into a reusable asset.
- Create Direct Upload Batch Requests up to 100 presigned upload URLs in one call.
- Complete Direct Upload Batch Finalizes every uploaded file in a direct-upload batch.
- Get Asset Batch Returns aggregate status for an asset upload batch plus member items.
Webhooks
7- List Webhook Endpoints Lists registered endpoints with their URLs, events, and status.
- Create Webhook Endpoint Registers an HTTPS URL to receive HeyGen event notifications like video completions and translation results. The step that lets renders report back to FlowRunner.
- Update Webhook Endpoint Updates an endpoint's URL or subscribed events.
- Delete Webhook Endpoint Permanently removes an endpoint. Reviewed before running, since it silences completion reporting.
- Rotate Webhook Secret Generates a new signing secret and invalidates the old one. Used in scheduled credential hygiene.
- List Webhook Event Types Returns all available event types with descriptions.
- List Webhook Events Returns a history of delivered events, filterable by type or entity.
Account and Brand
3- Get Current User Returns the authenticated profile plus remaining credits and billing details. The balance check before any large batch is submitted.
- List Brand Kits Returns the workspace's brand kits.
- List Brand Glossaries Lists brand glossaries, the custom term mappings that keep translations on-vocabulary. Used in the proofread comparison step.
Workflows
4- List Workflows Lists available HeyGen workflow types.
- Execute Workflow Submits a single HeyGen workflow for execution.
- Execute Workflow Graph Submits multiple workflows as a DAG so one workflow's outputs feed another.
- Get Workflow Execution Polls a workflow execution's status and output.
Frequently Asked Questions
What can FlowRunner do with HeyGen?
FlowRunner agents can run Create Avatar Video, Create Avatar Video and Wait, and Create Video from Image in HeyGen, plus 95 more actions.
Does connecting HeyGen to FlowRunner require OAuth?
No. HeyGen connects to FlowRunner with an API key, no OAuth flow required.
Can HeyGen trigger a FlowRunner workflow automatically?
HeyGen doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with HeyGen
$100 in credits. No card required. Connect in minutes.