How to turn Slack chatter into HeyGen videos with Claude Cowork
- Source
- x.com
- Author
- HeyGen
- Date
Context
HeyGen's workflow guide describes using Anthropic's Claude (in its Cowork mode) to read a team's Slack messages, propose three video topics and draft a script for the one the user picks. HeyGen then renders the script with the user's avatar and voice through its connector, the standard interface that lets Claude call HeyGen's video tool.
The guide keeps choosing a story and approving generation as separate steps, and says only the generate step spends HeyGen credits. That gives the user a chance to correct the script before rendering. It also recommends keeping that checkpoint when the preparation is automated, because a scheduled job can read half-finished threads, and a discussion still in progress can read like a settled decision.
- HeyGen's guide says to switch on only the connectors a job needs, because every enabled connector's tool list sits in Claude's and more tools make wrong tool picks more likely.
- The workflow keeps rules in a project folder instructions file that Cowork reads before each run, since Cowork does not carry memory between sessions on its own.
- For timing, the guide recommends word counts over seconds: avatars read about 2.5 to 3 words per second, so a 25-second update is roughly 65 to 75 words.
- Rendering is kept as a separate manual message because it is the only step that spends HeyGen credits; scheduled runs should automate the pitch, never the render.
- Per the guide, Claude reaches custom connectors from Anthropic's servers rather than the user's laptop, so a tool reachable only on an internal network will not connect.
- MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
Checking sign-in…
Loading comments…