Early-stage venture · San Francisco · Since 2016
We back founders before the market agrees.
First checks at pre-seed and seed, roughly twelve a year. Small enough that every company gets a partner who knows the details, patient enough to sit through the years when nothing looks like progress.
- First check written
- 2016
- Companies backed
- 180
- Combined portfolio value
- $4.8B
- Countries
- 41
Two partners, one fund, a lot of conviction
Roughly twelve a year, on purpose
Marked at last priced round
We have never required relocation
What we are looking for
Four places we spend most of our time
We are generalists by charter and specific in practice. These are the areas where we have done the work, made the mistakes, and can be useful on day one.
- 01
Software that does the work
Models became useful faster than companies became ready. The interesting businesses now are the ones absorbing real workflows, not the ones adding a chat box to an old product.
- 02
Infrastructure under the boom
Every platform shift makes a handful of boring, essential companies. Evaluation, data movement, permissioning, cost control. We look for the layer everyone will need and nobody wants to build.
- 03
The industries nobody demos
Logistics, insurance, clinical operations, energy markets. Slow to sell into, hard to displace, enormous once you are in. These take a decade, which is the point.
- 04
Founders with an unfair reason
The best pitch we hear is usually a small, specific thing the founder knows that we do not. Domain scar tissue beats a polished narrative every time.
Portfolio
180 companies, and counting
Most of them we met before anyone else was paying attention. A few are listed here.
- AI & MLNimbus AIAgents that run back-office operations end to end.Series B · First invested 2019
- FintechLedgerlyTreasury software for companies holding cash in six currencies.Series A · First invested 2020
- HealthcareHelix BioDesigned proteins for diseases too rare to attract pharma.Series B · First invested 2018
After the wire
The money is the easy part.
Every firm says it is helpful. What that means in practice is narrower and more boring than it sounds, so here is the actual list.
Recruiting
The first ten hires decide the company. We keep a live bench of operators and run your search alongside you rather than sending a list.
Customers
Introductions to buyers who have budget and a reason to answer, made by the partner who knows them. We do not spray your deck across a network.
The next round
We build the story, pressure-test the numbers, and tell you honestly when the round is not ready. That last part saves companies.
The quiet months
Most of the work happens when nothing is being announced. A call on a Sunday about a co-founder problem is the job, not a favor.
From founders
What they say when we are not in the room
“They told us to kill two of our three products in the first month. It was the correct call and nobody else was willing to say it out loud.”
“Our lead investor for the Series A came from a single introduction, but the more useful thing was the two months of honest feedback before we went out.”
“I called on a Sunday about a co-founder situation. Somebody picked up, and then stayed on the phone for an hour.”
Accelerator · Summer 2026
Twelve weeks, sixteen companies, one partner in the room.
$500K on a post-money SAFE, 7% target. No curriculum, no guest speakers, no demo day theatre. Applications for Summer 2026 close March 15, 2026.
- Investment
- $500K
- Ownership
- 7%
- Cohort
- 16
- Weeks
- 12
Insights
What we are thinking about
- PlaybookThe first ten hires decide the companyNobody believes this at seed stage. Then the eleventh hire arrives and inherits a culture that was set months earlier by people who were not thinking about culture at all.March 4, 2026
- FirmWhat we look for, written plainlyMost investment criteria are written to sound impressive. Here is ours, including the parts that make us look narrow.February 18, 2026
- PlaybookPricing before you have product-market fitCharging early is the fastest honest signal available to a young company. Most founders delay it for reasons that feel principled and are not.January 29, 2026
Tell us what you are building
One partner reads every application. You will hear back either way, usually within a week.