SOON COMING 02 / THE LAB
← BACK TO LAB
01WORKING PROTOTYPEPILOT TESTING WITH CLIENT DATA

AI MARKETING
DEPARTMENT

Can an AI team help a marketing agency see problems before the client does?

Agencies sit on huge amounts of useful information: campaign performance, analytics, creative, search data, client requests and reporting.

Most of it lives in different places, gets reviewed at different times and often only becomes important after something has already gone wrong.

I’m building an AI Marketing Department to test a different model.

One question. Six AI roles. One decision layer.

At the centre is an AI CMO. It receives the business problem, decides which specialist AI agents need to investigate it, briefs them, compares their findings, challenges conflicting recommendations and returns the few priorities that actually matter.

The initial team brings together the AI CMO with Performance, Growth, Creative, Content & Lifecycle and SEO AEO specialist agents. They work from a shared Client Brain containing the account’s objectives, performance context, decisions, learnings and customer data, rather than starting from scratch every time.

  1. 01The agency asksA real business problem enters the system.
  2. 02The AI CMO frames the problemIt decides what needs investigating and who should do it.
  3. 03Specialist AI agents investigateEach agent examines the evidence through its own discipline.
  4. 04The AI CMO challenges and synthesisesConflicts are tested and weak recommendations are removed.
  5. 05The agency gets clear prioritiesThe output focuses attention on the moves that matter.

Can it analyse the same evidence as an experienced agency team and produce a diagnosis they would actually find useful?

The first pilot is deliberately constrained: one client, controlled and masked data, and manual inputs only. No live APIs. No MCP connections. No access to client accounts. No autonomous actions.

Eventually, the idea becomes less like a collection of AI agents and more like an operating layer for an agency.

It could monitor account health, surface risks, prepare reporting and client meetings, track commitments, identify opportunities, coordinate specialist work and give agency leadership a daily view of where attention is actually needed.

Not another dashboard to check.
Something that checks the dashboards for you.

Explore a problem with me
G GROUND