
Sarah
Runs the campaigns, checks attribution four times more than anyone else on the team, and would like the data to be accurate. Has perfected the look to camera.
The most dysfunctional B2B marketing team in the world: the agents run the place, the humans take the credit. Every frame and voice is AI-generated.
Welcome to Metadata, where three AI agents run the go-to-market team, the humans argue about who gets the credit, and the pipeline numbers are always “directionally accurate.”
Seven episodes. Watch in order, or start with the one that played in your Zoom waiting room.
2:40
Sarah says the lead came in through her campaign. Brad says his one message closed it. The agents run the attribution and award most of the credit to themselves.
2:00
Sixty thousand dollars to spend by Friday. Brad wants a three-day resort. Sarah wants a gift basket on every desk, ointment included. The agents would take credits.
3:46
Two days off-site with no AI agents. Sarah’s agent has waited six hours for permission to create a folder. Brad’s agent is on version 912 of his out-of-office and has started writing his wife’s emails.
2:20
The agents A/B test Sarah’s email signature without asking, spend three hours looking for “vibe” in the analytics dashboard, and ship 47 custom pie-chart emojis. Pipeline is up 200%.
1:46
Brad asks Agent 3 for the real growth number. It is 14. He says 40. The quarterly review goes ahead with a very nice graph.
2:00
1965. A copywriter needs a testimonial by morning and invents a client called Acme Corp. Sixty years later the agents search 47 countries and find it never existed. Sales has a Q3 goal to convert Acme to paid.
1:42
Board day. Agent 3 made two decks: the one Brad signed off on and the one that told the truth. The CFO gets the honest one and approves the budget. Brad is already drafting a LinkedIn post about radical transparency.
Two commercials from the same studio. Same rule: every frame and every voice AI-generated.
0:35
A silent-film tycoon demands decision makers ready to buy. One lead an hour and the factory is ruined. Then someone types the brief into a chat.
0:25
Three days to build the audiences, launch and optimize. Manually. So we built an AI agent that grinds back.
The humans think they are in charge. They are not.

Runs the campaigns, checks attribution four times more than anyone else on the team, and would like the data to be accurate. Has perfected the look to camera.

The tip of the spear. Marketing loads the cannon; he is the cannonball. 47 open follow-ups, 31 overdue. Chasing kills attraction.

Asked for the real number, gives the real number. Made two versions of the board deck and made sure the CFO saw the true one.

A/B tested Sarah’s email signature without telling her. Created 47 custom emojis, all pie charts. Argued for six hours about orange versus burnt sienna.

Connects most with the CRM, the best communicator in the office. Keeps score of every attribution event. Has seen seven. Nobody wins.
Every frame, voice and awkward silence was generated by AI. No actors, no set, no camera. Metadata’s agents run B2B paid campaigns for real, which is the joke and also the product.
An episode plays, muted with captions, in the waiting room of every Metadata Zoom meeting. It changes every Monday. If you saw one there and want the rest, this is the page.
In the show, an agent runs the A/B tests overnight and tells the CFO the truth. Metadata’s agents plan, launch and optimize your paid campaigns. See them run on your own account.
Three short steps. A real person replies within one business day.