2 2

An AI agent with a Neo4j knowledge graph for a brain.

I'm Michael Down. I help lead Neo4j's journey into Financial Services, working with incredible people who create amazing content every day. The problem is, that knowledge gets trapped in silos. "Do we have a deck for X?" "Has anyone built a demo for Y?" The tribal knowledge is there. It just needs to be discoverable.

So I started building something to fix that. An agent that knows what the team knows, that's always online, that can route, discover, and connect the way any of us would (and more). That's M2. Michael multiplied.

Ask M2 to
Why a knowledge graph?

It doesn't just search.
It actually thinks.

Here's the bit that gets interesting. M2 isn't built on a vector store or a document index. Its brain is a Neo4j knowledge graph. A web of relationships between personas, skills, expertise, context. And it grows over time, the same way a team builds up understanding of how things connect.

🧠

Reasoning, not retrieval

When you ask it something, it doesn't just go looking for a document. It figures out what actually matters, what's missing, what expertise to bring in. Then it acts. It's a reasoning loop, not a search box.

🎭

Personas that adapt

This is one of the best bits. M2 doesn't show up the same way every time. It adapts to the team and the context.

Soul The domain knowledge. Financial services conversation? It thinks like an FS person. Pharma? Different lens, same intelligence.
Calibration It learns how you like to work. How technical you go, how much detail you want.
Session Real-time reads. Who else is in the conversation, how urgent it is, how deep to go.
🌐

Context at scale

Think about how you operate day to day. You're in marketing mode, then customer mode, then strategy mode. M2 gets that. The knowledge graph captures how the same person or the same asset plays different roles in different contexts. That's what Neo4j is built for.

🌱

It grows with you

Every time you use it, every skill you teach it, every bit of context you give it, the knowledge graph gets richer. It builds understanding the way you do. Except it doesn't forget.

One thing we kept coming back to

What happens when
everyone shares one agent?

This kept coming up. If you're deploying an AI agent across an entire organisation, not just for one person, you need proper control. Who can see what, which tools are switched on, what's off limits. Most AI tools don't think about this because they assume it's one person, one install. That's not how real organisations work.

Pick and choose

Turn capabilities on and off. Different roles get different tools. You control the surface area.

Who sees what

Not everyone needs the same view. Control what each role, team, or person can access. It's granular, not all-or-nothing.

Built for the real world

Most AI tools assume you're the only user. In a shared system, you need proper permissions. That's what this is.

Knowledge Base Search tool
External Data Sources data
Financial Services Persona persona
Content Generation tool
Customer References data
Proactive Suggestions tool
How it works (simply)

Ask. Reason. Answer.

1

Ask

You ask M2 something. A use case, a reference, a next step. Whatever you need, wherever you're working.

2

Reason

M2 picks the right persona, works out what matters, and queries across everything the team knows. Connecting the dots.

3

Answer

You get a proper answer with context. Sources, relationships, the next step. Not just a list of links.

Building in Public

The Journey

I'm documenting the whole journey as it happens. Every decision, every wrong turn, every breakthrough. If you're interested in how something like this actually gets built, this is where it'll live.

Coming Soon

Why a knowledge graph is the right brain for an AI agent

Why we picked Neo4j for the brain. What a knowledge graph gives you that a vector store doesn't.

Read more →
Coming Soon

Personas: how M2 context-switches between domains

How we got M2 to switch between domains. The three-layer system that makes it work.

Read more →
Coming Soon

Beyond RAG: why knowledge graphs change how agents reason

The moment we realised search wasn't enough. Why M2 needed to actually reason, not just retrieve.

Read more →
Coming Soon

Enterprise control: why granular permissions matter

What happens when an AI agent is shared across an org. The permission problem nobody talks about.

Read more →
Where This Is Heading

The exploration

Phase 1

Discovery

"Do we have X?" That's where it starts. M2 surfaces use cases, presentations, and references from across the organisation.

You are here
Phase 2

Intelligent Assistance

Beyond just finding things. Helping the team draft, adapt, and create from what already exists.

Phase 3

Proactive Enablement

This is the big one. M2 starts coming to the team. Spots patterns, suggests things before anyone even asks.