Make videos for your help articles with Auto Capture
Ask Claude, connected to Tella’s MCP server, to find help articles with no video and use Auto Capture to record each one from the article’s own steps.
Auto Capture is in a closed beta and isn’t available to every workspace yet. We’d love your feedback: email support@tella.com.
Ask an AI app like Claude, connected to the Tella MCP server, to find help articles that have no video and record one for each: it reads the article’s written steps, passes them to Auto Capture, and checks the finished video against the article. Nobody has to sit down and record.
In the video above, Claude audits Tella’s own help center, finds 179 of 252 articles with no video, picks the most-read one (“Add zooms and screen effects”), and records a 26-second demo of adding a zoom in the Tella editor straight from the article’s instructions.
An AI app connected to Tella. Claude, ChatGPT, Claude Code, Cursor or another client with the Tella MCP server. See Connect an AI assistant to Tella.
A saved login for your product, if the steps happen behind a sign-in. Add a demo account under Settings → Credentials, with Use on set to your app’s site. See Let Auto Capture sign in to your product.
Your help center’s address, such as https://example.com/help, so the AI can read the articles.
Ask the AI to record videos for your help articles
Paste a prompt like this into a chat with the Tella connector turned on, with your own addresses and login name:
Audit the help center at https://example.com/help and list the articlesthat don't have an embedded video. Pick the most-read one, then use TellaAuto Capture to record a video that follows the article's written steps onhttps://app.example.com, signed in with the "Demo account" credential.When it's delivered, check the video matches the article step by step andtell me anywhere it doesn't. Then put it in a new Tella video named afterthe article.
The AI then:
Reads your help center and lists the articles with no video.
Starts an Auto Capture take, passing the article’s steps as the instructions. The capture agent signs in off camera, writes a script, checks it runs, then records it.
When the take is delivered, compares what the video shows with what the article says.
Puts the take in a new Tella video, ready to trim, narrate and share.
Ask for one take per article. For a batch, ask it to work through the list one article at a time.
Name a saved login in your prompt, never a password. Auto Capture only signs in with logins saved under Settings → Credentials.
If the capture agent can’t follow an article’s steps, a reader probably can’t either. When a take ends with a question, or the check finds the video and the article disagree, that usually means a step is missing, out of order, or names a button that has since changed. Fix the article and ask for a new take.
When your product changes, ask for a new take of the articles it affects instead of re-recording by hand. If you already have an automation that updates help articles when a change ships, add a step that asks your AI app to record a fresh Auto Capture video for each article it edits.
Open the new video in Tella, trim it or add narration if you like, then share it and paste its embed code into the article. See Customize embed settings, or Embed Tella videos in Mintlify if your help center runs on Mintlify.
