Replace Serendipity with Structure
A Better Way to Build Startups from Research
When I tell people I’m a Venture Builder at MIT, they often imagine I spend my days mentoring startups or roaming the halls of MIT labs looking for promising intellectual property.
But the reality is much more methodical.
What Venture Builders actually do is build structure around serendipity.
Because if there’s one thing I’ve learned in this role, it’s this:
Breakthrough ideas don’t launch themselves.
And brilliant science doesn’t magically become a business — unless someone is building the scaffolding beneath it.
So here’s a peek inside my typical flow. Not the polished success stories, but the workflows, tools, and processes that help us turn early insights into real ventures.
Map the Tech Terrain
The Venture Builder process starts with a core activity: tech scouting.
I attend research seminars, read through preprints and recent publications, and take postdocs or graduate students out for coffee to ask one deceptively simple question:
“What have you built lately that the world hasn’t seen yet?”
Sometimes it’s a new way of using an existing sensor. Sometimes it’s an intriguing algorithm. Sometimes it’s a whole system.
Each of these goes into our knowledge system (we use Obsidian at Proto Ventures). For every capability, I log:
TRL (Technology Readiness Level)
IP status
Key risks
Potential applications
People involved
Over time, this becomes a living map of the technical frontier inside the lab.
Talk to the Outside World
The next crucial step is market discovery.
That means 30-minute calls with operators from startups and big companies alike, as well as investors, CTOs, and industry consultants.
My goal isn’t to pitch. It’s to listen:
What problems are getting worse?
What would they fund if someone built it?
What technology are they looking for but can’t find?
I log each market need as a structured note:
Severity (1–10)
Existing solutions
Total addressable market
Known startups working on it
Signals from public funding (ARPA-E, SBIRs, etc.)
Over time, this becomes a mirror image of the map—a set of “pull” signals to complement our “push” technologies.
Connect the Dots
From there, the name of the game is pattern matching. I cross-reference capabilities from the lab with needs from the market.
This is where we generate venture ideas—not based on what one lab is building, but on what the world actually needs that this lab could uniquely deliver.
We log each venture idea with:
A one-paragraph description
Associated capabilities and needs
Early hypotheses about product, market, and risks
Notes on team formation potential
This is where ideas start to feel real—and also where most ideas die. We’re ruthless at this stage, because focus matters.
Zoom In
For the top 3–5 ideas in our current channel (e.g. fusion, geothermal, hydrogen), we go deeper to explore the idea.
That means:
De-risking technical assumptions (talking with the PI or doing quick experiments)
Modeling first product concepts
Sketching business models and cost curves
Developing pitch decks or mockups
We often run "mini build sprints"—1–2 week collaborations with Fellows or outside advisors to push an idea forward fast.
This is where we find out: is this something someone could build, fund, and lead?
Evaluate
Next is sharing and stress-testing.
At MIT Proto Ventures, we hold internal “stage gate reviews” where we present venture ideas and get feedback from the studio team, faculty leads, and domain advisors.
We ask ourselves:
Is this idea non-obvious and defensible?
Is there a real market pull?
Are we uniquely positioned to build it?
Who might lead this? (Fellow, researcher, VB?)
This structure creates accountability—but also leaves room for creative leaps.
What Structure Really Does
To be clear, this isn’t rigid. There’s no “checklist” for deep tech innovation.
But structure gives us rhythm. It makes sure we don’t get stuck chasing one idea too long—or miss a diamond in the rough.
It helps us turn a hallway conversation into a proto-venture in a matter of weeks.
As MIT Proto Ventures faculty directory Fiona Murray writes:
There is a huge opportunity cost to letting great research sit idle. The venture studio model makes innovation systemic rather than accidental.
Next in our series: “Founding Teams Don’t Form Themselves—How Studios Orchestrate Entrepreneurial Chemistry.”