HeyGen + YouTube: Campaign Video Rendered, Reviewed, Then Published

How do you connect HeyGen to YouTube?

A HeyGen render completing, whether from Create Video Batch, Generate Video from Template, or Create Video Translation, triggers an Upload Video call on YouTube that publishes the finished file to your channel as a private draft with Set Video Thumbnail applied. FlowRunner is a visual AI-agent orchestration platform where automations run autonomously and pause for human judgment on the steps that carry real consequence. The same connection can run as an AI agent that reasons about which cut is ready, checks it against the brand glossary, and invokes a human-in-loop step before a video with a cloned voice or a public audience goes live. That upgrade is the difference between a sync tool and an orchestration layer.

The problem it solves

A marketing ops lead running a campaign with five language variants ends up managing two separate manual chains. The first is production: submitting each HeyGen render one at a time, checking status by hand, downloading files, and tracking which variant is done in a spreadsheet. The second is publishing: uploading each finished file to YouTube, writing the title and description, setting a thumbnail, and remembering which cuts still need a human to actually look at them before they go live. Neither chain talks to the other, so the handoff between “video is rendered” and “video is on the channel” is a person copying a file from one tool to the next.

The risk compounds with scale. A campaign with 20 or 30 template variants means 20 or 30 manual uploads, each one an opportunity for the wrong cut, the wrong thumbnail, or an unreviewed translation to reach an audience. Errors caught after a video goes public cost more than errors caught before, and a producer checking render status one video at a time doesn’t scale past a handful of variants without falling behind.

How it works: the connection

The flow starts on the production side. The agent reads a template’s variable schema with Get Template, fills it from the campaign sheet, and submits up to 100 renders in a single Create Video Batch call. Instead of checking each video individually, it tracks the run with Get Video Batch and Get Bulk Video Statuses, which return statuses for up to 100 videos in one request.

As each cut finishes, Save Video to File Storage pulls it into FlowRunner’s storage and returns a stable URL. For the hero cut destined for the channel, the agent calls Upload Video on YouTube to publish the file as a private draft, then Set Video Thumbnail applies the campaign artwork and Update Video fills in the title, description, and tags. The private draft link goes into the content review channel, matching the same pattern YouTube’s own connector uses for any upload: the mechanical work is automated, and the go-live decision waits for a person.

Can an AI agent run it? (and why a human stays in the loop)

The agent does not just move files from Create Video Batch to Upload Video in a fixed order. It reads the batch results, reasons about what it finds, and picks its next action from the tools available to it, including a human-in-loop step it can call exactly like any other action.

Here is a real decision moment. The campaign includes a German cut that uses a cloned voice of the company’s spokesperson, one of several variants Create Video Translation produced with voice cloning enabled. Before that cut renders, the agent opens a Create Proofread Session, pulls the draft subtitles, and compares them against the terms in the brand glossary. It finds a product name translated inconsistently with the glossary and a line that reads differently in tone from the source script. Rather than proceeding, the agent packages the context, original and edited SRT links, the specific lines it flagged, and a summary of why, and posts it to the regional reviewer in Slack: “The German track for the launch video is ready. Review the subtitles here, original and editable SRT attached. Rendering uses your spokesperson’s cloned voice and consumes render credits; reply approve when the transcript is right.” The reviewer corrects two lines and approves. Only then does Generate Video from Proofread spend the render credit.

The second gate sits later, at the YouTube boundary. Once the hero cut is uploaded as a private draft, the agent does not flip it public on its own. It posts the preview link to the content review channel with a plain question: approve to publish, or send it back? A channel owner watches the actual video, not just a status report, and makes the call. This is not a hardcoded rule triggered by a threshold. It is the agent recognizing that a cloned voice and a public channel are both consequential enough to warrant a person’s sign-off, and routing to that person with everything they need to decide quickly. This is not a sync tool. The agent knows when to stop and ask.

FlowRunner vs Zapier

Zapier is a genuinely fast way to wire a HeyGen webhook to a YouTube upload for a single video, and its editor is approachable for a non-technical marketing team building a simple two-step Zap. For a one-off “when a video finishes, upload it” connection, that simplicity is real and worth acknowledging.

Where it runs into limits is exactly the scenario above: a batch of variants, a translation review gate, and a publish approval that has to hold specific videos back without stalling the rest of the campaign. Zapier’s model is built around linear steps and paid Filter or Path add-ons, not a native review-and-resume pattern.

Capability	FlowRunner	Zapier
Human-in-the-loop	Native tool the agent invokes mid-workflow, with full context and a resume step	No built-in approval pattern; typically bolted on with a separate app or manual check
Users included	Unlimited on every tier	Priced per user on most plans
AI provider	BYOK, bring your own model and provider	Zapier-hosted AI credits, usage-metered
Batch handling	Agent reasons across a full batch and holds specific items back	Each run is largely independent; batch logic needs custom workarounds
Pricing model	Transparent workflow-based tiers with a fixed execution ceiling	Per-task pricing that scales unpredictably with campaign volume

Before and after

Metric	Before	After
Campaign variant production	Each language and clip is its own manual project	Create Video Batch and a shared template produce all variants as one reviewed run
Translation quality control	Errors surface after publish, when every fix means a full re-render	A reviewer corrects subtitles as text before Generate Video from Proofread spends a render credit
Upload workflow	Uploading, titling, tagging, and thumbnailing is a hand sequence per video	The agent uploads as a private draft with metadata and thumbnail already set
Publish decision	Publish depends on whoever remembers to check the queue	A channel owner approves from a Slack preview link before the video goes public
Status tracking	A producer checks render status one video at a time	Get Bulk Video Statuses reports up to 100 renders in a single call

What you can build

Template campaign to channel launch. Fill a HeyGen template’s variables from a campaign sheet, submit the batch with Create Video Batch, and push the approved hero cut to YouTube with Upload Video and a matching thumbnail.

Localized channel library. Run Create Video Translation with voice cloning for each target market, route the proofread session to a regional reviewer, then upload the corrected SRT to YouTube with Upload Caption alongside the localized video.

Webinar to shorts pipeline. Submit a long recording to Create AI Clipping Job, let the content owner pick which clips represent the brand, and upload the approved clips to YouTube as private drafts pending a channel owner’s sign-off.

Weekly spokesperson update. Generate a recurring video from a template using a Digital Twin Avatar, hold it for the same reviewer who approved the original consent flow, and publish it to a standing playlist once approved.

Getting started

Connect HeyGen and YouTube in FlowRunner and build this as a plain connection or as an AI agent with the review gates described above. New accounts get a $100 credit on the Growth tier, roughly 67 days of typical use, with no credit card required. Start at flowrunner.ai or book a walkthrough at calendly.com/flowrunner/intro.