Skip to content

Eleven Agents, No Team

Bhanu Chawla, CMO at Frame, opens Frame Game, the retro control room he built for the eleven agents that run marketing at Frame with no team. They design the ads, write the copy, launch the Meta campaigns, debate ideas in a boardroom, and report back to him for approval.

Twenty to twenty-five times the output, on a $200 a month Claude subscription and a Mac Mini.

Published 2 September 2026 · 36 min · Agent Cribs EP.02

What happens in this episode.

Bhanu Chawla runs marketing at Frame with eleven AI agents and no team. In this episode of Agent Cribs he opens Frame Game, the retro control room he built for them on a Mac Mini. Each agent has one job: Aria runs the day as chief of staff, Pixel designs, Quill writes the copy, Tracer runs the paid ads. They pass work between each other, launch Meta campaigns, and only come to Bhanu for a yes.

The tour covers the design choice behind it, why several small agents beat one big one, how they talk to each other, and what happens when a campaign runs on its own. Bhanu shows the approval loop in detail: every ad, post and campaign comes back to him, and a rework is how the agent learns what to do differently next time. He also runs a multi-agent debate on camera, the boardroom where the agents argue an idea before anything gets built.

The whole setup runs on a Claude Max subscription, about $200 a month, with no API bill. Bhanu puts the gain at twenty to twenty-five times his previous output, and his line for the way he works with it has stuck: AI brings the speed, humans bring the taste. The last stretch of the episode is about onboarding a human into the system, building an interface that fits how you think, and how long it really takes to train an AI employee.

On screen

  • Frame Game, the retro control room over eleven agents, and the welcome-to-the-crib walkthrough.
  • Aria, Pixel, Quill and Tracer at work: chief of staff, design, copy and paid ads, each with one job.
  • A campaign triaged, built and launched on Meta, with the cost of running it on a $200 a month subscription.
  • The approval loop: what comes back to Bhanu, what a rework looks like, and what the agent keeps from it.
  • A multi-agent debate in the boardroom, and how a human gets onboarded into the system.

What to copy, and what not to.

  • 01

    Eleven agents, one job each. Aria runs the day as chief of staff, Pixel designs, Quill writes the copy, Tracer runs the paid ads. They pass work between them and only come to Bhanu for a yes.

  • 02

    He is the approval step, on purpose. Every ad, post and campaign comes back for a yes or a rework, and the rework is what the agent learns from for next time.

  • 03

    It all runs on a Claude Max subscription and a Mac Mini, about $200 a month with no API bill. His line for the whole thing: AI brings the speed, humans bring the taste.

The guest

Bhanu Chawla

CMO at Frame

Bhanu is the founding CMO at Frame, a settlement layer for global finance. He has spent 16 years in B2B SaaS marketing and started out as an ethical hacker in Delhi, which shows in how he builds his own tooling rather than waiting for someone to ship it.

The short version.

What is Frame Game?

A retro, game-styled control room Bhanu built on a Mac Mini to run his eleven agents. Each agent has a name and one job, and the interface shows what they are working on and what is waiting for his approval.

Which agents does Bhanu run?

Eleven Claude-based agents. The ones he shows on camera are Aria, the chief of staff who runs the day, Pixel for design, Quill for copy and Tracer for paid ads. They hand work to each other and report back to him.

What does it cost to run eleven agents?

About $200 a month. The whole setup runs on a Claude Max subscription and a Mac Mini, with no API bill.

Does Bhanu approve everything the agents make?

Yes, by design. Every ad, post and campaign comes back to him for a yes or a rework, and the rework is how the agent learns what to change next time.

How much more does he get done?

Bhanu puts it at twenty to twenty-five times his previous output, with no team. His line for the split of labour: AI brings the speed, humans bring the taste.

What is
Patricia?

Patricia is the AI coworker behind the show. It lives in Slack and Teams: you connect your tools once, then just ask, and Patricia picks up the busywork so you can focus on the parts that need a human. Collaboration, not replacement. That is the whole point of Patricia, and it is why this show is about how real people work alongside their agents.