TRAC.vc lets the numbers say yes.

A venture capitalist calls a founder who has never pitched him. The founder expects the usual exploratory conversation. But the caller has already reached a decision. Somewhere in a database, the company has crossed a threshold. Now comes the awkward part: will the founder let the investor in?

This is the inversion at the center of TRAC.vc. The firm says its algorithms make every investment decision. Its partners handle the conversations. The distinction sounds like a small organizational choice until you consider how much of venture capital revolves around the judgment of the person across the table.

The signal behind the signal

TRAC stands for Third Round Analytics Capital. Its current team page names four co-founders: managing partners Joe Aaron and Fred Campbell, data science head Steve Marek, and engineering head Scott Pyne. That combination matters. Someone has to define the question, someone has to build the model, and someone has to make the machinery useful.

The question TRAC asks is unusually indirect. Instead of beginning with a founder’s persuasiveness, it examines the investors who have already committed capital. Which individuals repeatedly found promising companies early? What happens when several of them back the same business? TRAC calls selected repeat performers SuperTRACers.

A famous firm’s logo is an imprecise answer to a precise question. A partnership contains different people with different records. TRAC’s hypothesis is that individual histories reveal something the brand obscures. Its system turns other investors’ past decisions into evidence for a new decision.

“Data is our investment Committee.”

TRAC’s statement of investment authority

There is an amusing consequence. Human judgment remains inside the process, embedded in the historical choices the model studies. TRAC relocates discretion: its own partners do not get the final selection vote. That makes the rules consequential, including when a partner likes a company the model rejects.

A yes still needs a sales pitch

Legacy, the male fertility company, illustrates the experience from the receiving end. Founder Khaled Kteily described an unsolicited approach from TRAC before he was actively raising his next round. “What impressed me most was the transparency and efficiency,” he said. The model’s decision gave the conversation a concrete starting point.

But a company attractive to TRAC can be attractive to many other investors. The scarce resource becomes space in the round. A modest check still needs an answer to the founder’s entirely reasonable question: what else do you bring?

TRAC’s answer is its Intelligence Dashboard. Portfolio founders can examine competitors, compare valuations and identify prospective investors. The same information that helps TRAC decide where to invest becomes something a founder can use after the money arrives. Capital buys the firm a stake; useful intelligence helps it earn the invitation.

Put today’s number first

The dashboard has its own small history of learning. In a 2023 account, Campbell explained that an earlier version emphasized a company’s future valuation after its next raise. The third version put current valuation on the home page, alongside market rating and monthly valuation history. Founders needed to know where they stood before deciding where to go.

That adjustment is more instructive than a grand claim about artificial intelligence. A prediction becomes useful when it meets a decision someone must make. Should we raise now? Which comparable deals support the asking price? Which investors have already backed a competitor?

For a founder, the useful exercise is to separate a valuation estimate from the argument behind it. A number becomes more persuasive when the comparison companies, timing and assumptions can be examined. Change those inputs and the answer may change too. The dashboard’s appeal is that it gives founders material for that discussion. A price suggested by software still has to survive a negotiation with someone writing a check.

The data work underneath is less glamorous. Similarweb’s case study describes the challenge of obtaining reliable, comprehensive inputs. Its contribution includes traffic, engagement and search signals. Integrating these with other sources gives the dashboard more context, though the relevance of any web metric still depends on the business being measured.

Correlation Ventures and SignalFire also put data at the center of their public positioning. TRAC’s distinguishing claim concerns authority: it says the algorithm selects every investment. “We use data” tells you little until you know who can ignore it.

Two pools of money. One long clock.

The public filings make the financial structure tangible. Fund I reported $31.175 million sold in its June 2022 amendment. Fund II reported $26.3 million sold in its July 2022 notice. These were contributions to investment vehicles, not sales revenue for a software business. Fund I’s disclosure also specifies an annual management fee linked to assets under management.

Consider the two sides of the transaction. Limited partners commit money to a fund; founders accept investment into a company. They are making different purchases, with different expectations. The founder gets capital and access to intelligence while giving investors an economic stake. The limited partner relies on the fund to turn those stakes into eventual proceeds. A useful dashboard can improve the first relationship well before the second relationship has delivered its verdict.

The firm’s website now reports more than $65 million invested across more than 120 companies. Those are measures of activity. Its predictions require another clock: whether companies eventually produce proceeds for investors. A follow-on round demonstrates continued financing; a valuation marks a price at a moment. Neither is interchangeable with returned cash.

In March 2026, Business Insider published another list of 30 potential unicorns selected using TRAC’s model. The stated one-in-five probability is a forecast. Keeping that distinction clear makes the experiment more interesting: it leaves something specific to measure later.

What a founder can borrow

You do not need TRAC’s models to copy three habits: keep public company records accurate, build a defensible set of valuation comparisons, and check an investor’s existing portfolio before sharing sensitive information. Each makes a fundraising conversation better informed.

There are limits to inference from traces. A company with little recorded activity offers less to measure. A business whose progress happens offline may produce misleading web signals. And a model can identify a desirable investment without obtaining access to it. TRAC’s story makes that last constraint wonderfully concrete. Once the numbers say yes, someone still has to pick up the phone.