One Google Doc Coordinates All 14 of My AI Agents

One Google Doc Coordinates All 14 of My AI Agents

I was on a call with Dieter a couple of months ago. He’s deep into AI agents… runs a bunch of them for his architecture visualization business. But he had a problem I recognized immediately.

“Every time my priorities change, I have to go into each agent and rewrite the prompt. It takes forever.”

I pulled up my screen and showed him something. A Google Doc. Nothing fancy. About 3 pages. I call it the shared brain.

The Problem With Multiple Agents

Here’s what happens when you start building out a real agent system. You get an email agent. A content agent. A meeting prep agent. A research agent. Maybe a CRM updater. Each one has its own prompt, its own instructions, its own idea of what matters.

And when your priorities shift… which happens every week if you’re running a business… those agents don’t know. Your email agent is still prioritizing responses to old threads. Your content agent is writing about topics you’ve moved past. Your meeting prep agent doesn’t know which projects are active and which are on hold.

You end up spending more time managing the agents than the agents save you. That’s backwards.

What the Shared Brain Actually Is

It’s a single Google Doc that every agent reads before it does anything. That’s it.

The doc contains:

• My current top 3 priorities (updated weekly)
• Active projects and their status
• KPIs each agent should optimize for
• Which types of tasks are urgent right now
• Any temporary rules or exceptions

So when I update the file to say “AI consulting implementation is the main priority this quarter,” every agent shifts behavior. Email drafts prioritize conversations with consulting clients. Content angles toward AI implementation topics. Meeting prep pulls context for those specific calls. CRM tagging flags consulting-related contacts.

One update. All agents aligned. Instantly.

This is what I think of as the agent-as-teammate model. The long-term goal isn’t a bunch of isolated bots following rigid scripts. It’s a team of agents that are aware of your priorities, your history, and your context. The shared brain is how they stay coordinated.

Why It’s Embarrassingly Simple

The thing that surprises everyone when I show this: it’s not a database. It’s not some fancy multi-agent coordination framework. It’s not RAG or vector embeddings or any of that.

It’s a Google Doc.

But it works for two reasons. First, every modern AI tool can read a Google Doc. Lindy connects to it natively. You can paste it into Claude or ChatGPT. It’s the most portable format there is.

Second, one source of truth beats distributed truth every time. When agents all read from the same file, they can’t get out of sync. When you update it, you update it once. Done.

I learned this from managing real teams, honestly. The best-run companies I’ve seen all have something like this… a one-page strategic priority doc that everyone references. The shared brain is just the AI version of that.

A Real Example

Here’s how it played out last month. I shifted my focus from workshop planning to consulting implementation. Took me about 5 minutes to update the shared brain.

Within the next cycle, my email agent started drafting more detailed responses to consulting prospects. My content agent shifted toward implementation-focused topics. My meeting prep agent started pulling relevant case studies before consulting calls.

I didn’t touch a single agent prompt. Just updated the doc.

Dieter’s reaction on the call: “Wait, that’s it? I’ve been spending 30 minutes rewriting prompts every time something changes.”

How to Build Your Own

If you’re running 2+ agents and they feel disconnected, here’s how to start:

1. Open a new Google Doc. Title it something like “Agent Priorities” or “Shared Brain.”
2. Write down your top 3 priorities for this week or quarter. Be specific. Not “grow the business” but “close 3 consulting deals by end of March.”
3. List your active projects with a one-line status for each.
4. Add any rules or preferences your agents should follow. “Always draft, never send.” “Prioritize Austin-based contacts.” Whatever matters right now.
5. Connect this doc to every agent you run. In Lindy, add it to the Knowledge Base. In other tools, paste it into the system prompt or reference it as a file.
6. Update it weekly. Takes 5 minutes during your weekly review.

That’s the whole system. No code. No database. Just a document that keeps everyone… human and AI… on the same page.

The Bigger Picture

The shared brain is actually the minimum viable version of something bigger. As your agent system grows, you can layer in per-agent role definitions, person-specific dossiers, and decision logs. But the shared brain is always the foundation.

Start with the text file. Get your agents reading from one source. Everything else builds on top of that.