AI coworkers
your company
owns.

The brief

An executive needed to build a go-to-market team to handle all of the company’s sales and marketing work, except meeting directly with buyers. That part had to be done by people.

Using CopilotKit OpenBot, the company made 20 AI agents, and one person built them all and runs them.

How the team is built

Each agent holds a real GTM role, such as growth strategist, ABM specialist, demand gen manager, enterprise SDR, RevOps engineer or pipeline strategist. Each one has a written job description and works in its own tools. Across the team, those come from seven products:

The agents picked their own names, each after a TV or film character who fits the job. That is how the demand gen manager became Don Draper, the program manager became Leslie Knope and the librarian became Rupert Giles.

One agent sits above the specialists. Nicky Fury, the orchestrator, takes direction from the executive, assigns each piece of work to the right specialist and coordinates the hand-offs between them.

How a cycle runs

Work starts with a message from the executive. It can be a single line, such as “Build pipeline for the AWS launch by Friday.” Nicky Fury works out who should do what and passes each part to a specialist, who does it in their own tools and hands the result on to whoever needs it next.

When the outreach is ready, the executive approves it before it goes out.

Every night, a QA analyst audits the team’s output. A program manager runs the work in two-week GTM cycles, and a librarian keeps the team’s shared memory, so lessons from one cycle carry into the next.

Who does what

The executive sets the strategy and makes the decisions only a person can make, and people still hold every meeting with buyers. The agents do the rest of the work, across strategy, marketing, sales, data and RevOps.