YouTube Discovery, Product Announcement

Now Available: Your AI Agent Can Read All of Public YouTube

YouTube is now available to your AI agent! The platform holds billions of videos worth of signal and helpful information, but is not agent friendly to access.

YouTube Discovery hands that signal straight to your AI agent. It is an agent tool on AgentPMT, the integration platform for AI agents, and any agent in your group can find it, read its schema, and start pulling data through AgentPMT's dynamic MCP, the tool-delivery layer that loads a tool only when it is actually used so it never sits in context running up the bill. No account linking, no keys to rotate, no quota spreadsheet.

Point an agent at it and it searches the public side of YouTube the way a researcher would: videos, channels, and playlists filtered by keyword, topic, publish date, language, region, duration, license, and livestream status. It pulls rich metadata and performance statistics, view counts, likes, comments, subscriber totals, tags, and topics for up to fifty videos in a single call. It profiles any channel by handle or ID, tracks what is trending in a given country and category, browses full playlists, and reads public comment threads so your agent can gauge audience sentiment and the questions that keep coming up. Everything comes back as clean, structured fields with ready-to-open links, so the agent reasons over the results immediately instead of scraping HTML. It runs pay-per-use with no required subscription, and a failed call refunds itself, so an autonomous agent that skims the trending charts once a day and one that pulls ten thousand video stats for a quarterly teardown each cost exactly what they use.

The real payoff shows up once you wire it into a workflow. A webhook fires the moment a competitor uploads, and your agent pulls the new video's statistics, weighs them against that channel's posting cadence, and drops a share-of-voice update into the team's channel before anyone opens a browser. Put it on a schedule instead and an agent sweeps the trending charts for your category every morning, cross-references the winning titles and tags against your own backlog, and hands back a ranked list of what to make next: SEO and keyword discovery driven off what is already earning views rather than a hunch.

Because it behaves like any other workflow node, YouTube Discovery composes with the rest of the catalog. Chain it into a content pipeline that researches a topic, drafts the script, and files the plan in a doc. Feed its channel statistics into an influencer-vetting flow before a partnership clears. Route trending comment threads into a media-monitoring workflow that flags sentiment swings and surfaces recurring complaints. The agent doing the work can be driven from Hermes, OpenClaw, Codex, Claude Code, or ChatGPT, or it can run fully autonomously with no local setup at all through the dynamic MCP. That is business workflow management where the audience research runs itself.

Most of the recent automation in media has pointed at generation, the tools that write, edit, and render. Discovery lagged, which never made much sense, since no amount of production polish rescues a video aimed at the wrong topic. Media AI solutions that can actually read the platform, benchmark engagement, and track sentiment close that blind spot, and the teams building on artificial intelligence in media and entertainment are the ones who quit treating audience research as a quarterly scramble and let their agents do it continuously. The distance between them and the teams still copying video URLs into a spreadsheet is widening right now.

The public side of YouTube has always been readable. What is new is an agent that reads all of it on demand and acts on what it finds. Connect one to YouTube Discovery and put your video research on autopilot.