Network intelligence
Building network intelligence at ecosystem scale
What if you could query a network like a database?
I've been obsessing over this question for the last couple years while building MaRS Discovery District's platform. We're trying to connect 1200+ startups with exactly the right expertise, investors, and buyers they need.
The first thing we built seemed simple enough: an advisor search using OpenAI. A founder type’s "I need help with reimbursement for our medical device in Germany" and boom, ranked matches based on actual fit, availability, domain depth. But after processing thousands of these queries, I realized we'd stumbled into something bigger. We weren't just building a better search tool. We were creating network intelligence.
The limits of organic networks
I've personally hit Dunbar's limit. Yes, you really can only maintain ~150 meaningful relationships. And honestly, I can maybe track 5-7 things about each person. But here's the thing: ecosystems need expertise across dimensions no one person can hold. Operational experience, geography, market timing, fundraising stage, regulatory knowledge, customer connections. It's too much.
This is a hidden expertise problem. The person you need is probably two degrees away. But finding them? That's months of events and reaching out trying to set up perfect meetings.
Everything changes when you can literally type: "Series A healthtech founder expanding to Germany needs regulatory expertise."
From serendipity to systematic
Here's what's wild, we're starting to rewire and code how networks form.
The platform is starting to track live data: who just raised, who's expanding where, who's hitting regulatory walls. Real-time context. No more archaeology through stale LinkedIn profiles. We're getting ecosystem telemetry.
We're engineering the serendipity. That perfect investor you'd randomly meet at next year's Hack the North? Found today. The advisor who solved your exact problem at another startup? Surfaced now.
It's like having a time machine for connections.
Networks as infrastructure
This goes way beyond better search. Networks themselves are becoming programmable. Relationships are queryable. Expertise is discoverable at scale. The API literally becomes how you interface with collective intelligence.
We're augmenting human capacity to see patterns across thousands of dimensions at once. No human can do this alone.
A new primitive
What we're really doing is evolving from managing contacts to comprehending entire ecosystems. Understanding the hidden expertise graph that connects everyone.
Here's my bet: network intelligence will become as fundamental to organizing human knowledge as databases became to organizing information.
What we're building at MaRS is the beginning. We're making the implicit explicit, turning ecosystem expertise into something you can actually query.
Networks aren't just going digital. They're becoming computable.