Networks built to order

Nihar treats deal flow as a byproduct of the network-building and portfolio support that fill BTV’s days. “We treat sourcing as a second-order derivative of all of that,” he says. A tended network throws off deal flow on its own: a founder they’ve backed makes an introduction, or an operator they met for a portfolio search turns out to be starting something.

The large, multistage firms can miss early rounds but still find ways to invest later. BTV can’t. It has to win the founders it wants, who usually have other options, and give each company enough hands-on help to change its trajectory. 

Winning the right to invest in the best founders and supporting them to success often comes down to who the firm knows. Most VCs work the networks they’ve inherited, but BTV remains conscious of the value of expanding them. “Existing networks matter and compound over time. But you need to be smart about how you grow them. Unless you’re constantly refreshing your connections, your networks become stale, and you’re less valuable to your portfolio.”

Harmonic is how BTV drives that expansion. Scout, Harmonic’s AI research agent, takes a plain-language description of a profile (product owners at vertical AI companies in a given city, for example) and returns the people who fit it, drawn from a database of 35 million companies and 195 million professionals that Harmonic rebuilds continuously, so a job change or a new role shows up as it happens. Nihar can describe the network he wants and get it back as real names, a job that would otherwise take days by hand. Those names feed the portfolio: a customer to introduce, a candidate for a hard-to-fill role, an operator to reach before anyone else does, an invitee for the next event BTV convenes.

One way BTV activates a network is by gathering it in person. This past June, it ran a Vertical AI Summit in New York in partnership with fellow Harmonic customers, Ramp and Greylock, built around a specific set of conversations: vertical AI pricing, go-to-market, product, and economics. The guests were operators and founders from companies like Legora, Rogo, Profound, and Basis. The guest list began in Harmonic. “We asked Scout for the vertical operators in New York, the engineers and business-operations people at leading vertical AI companies,” Nihar says. “Harmonic was terrific at building that list.” The event space held 150. The thousand-plus people who signed up are now a network BTV can see and reach, much of it new to the firm.

‍

The entire firm’s network, one query away

A seed fund’s most valuable gift to a portfolio company is often a warm introduction to a customer it needs or to the investor who leads its next round. Making the right intro is how a five-person fund out-supports much larger ones. Each member of BTV brings their own, uniquely valuable connections to the table; Network Mapping pulls them together. The team synced the team’s LinkedIn connections and emails into Harmonic, which projects the firm’s collective relationships onto its startup graph and surfaces the best path into any company or person, whoever on the team holds it.

The Harmonic MCP turns that mapped network into something Nihar can query in conversation. The MCP connects Harmonic’s company and people data, and BTV’s own mapped network, directly to Claude, so a question asked in plain language comes back with an answer inside his workspace. “We get portfolio updates every month, and one of them might ask for intros to a few target customers,” Nihar explains. “I can port that email into Claude and ask who I know at those companies.” And because the whole team’s network is mapped, the answer extends beyond Nihar’s own contacts: if the warm path runs through a colleague, the MCP surfaces it, and Nihar hands the intro to the partner who holds the relationship. The monthly update stops being a request BTV can only partly answer and becomes a lookup against everything the firm knows.

‍

Filling founders’ most difficult open roles

Portfolio companies come to BTV with the hires they can’t make on their own. An early-stage team has no in-house sourcer, and the person it’s looking for is usually already employed, not actively looking, and working somewhere the founder has never heard of. Yoni Lateiner, BTV’s Head of Talent, is its most active Harmonic user, and hiring for the portfolio is most of what he does with it. “You can’t post the job and wait for the right person to answer,” Yoni explains. “The person our portfolio company wants is head-down somewhere, and likely not even on the radar of the company that needs her. We go find her.”

A founder will ask for something specific: an engineer with specialized fintech experience, building end-to-end systems and UX, who’s experienced fast growth and is based in San Francisco. The title and the city are easy enough to filter for, but the experience is the tricky part: almost no one has the detailed company context required, and the ones who do rarely spell it out on a profile. Harmonic finds them anyway. It knows which companies do compliance work, how they’re traction has evolved, and everyone who’s worked at them, so a qualified engineer surfaces through her employer, even when her own profile never uses the word. The platform’s coverage surfaces the two-person startup as readily as it does the household name, and People Search sets the filters (current or prior company, funding stage, job title, location, industry, technology, and more) to return everyone who fits. “That search used to be the hardest part, weeks of time spent researching companies to understand their particular work and domains and mapping their talent to the teams doing that work,” Yoni says. “Now the whole field’s in front of me, and the job is choosing between them, instead of wondering who I missed.”

‍

Why Harmonic, and not a general model

Every search BTV runs in Harmonic could in principle run against a general model or LinkedIn instead, and Nihar has thought about why BTV doesn’t work this way. A model on its own has no current, trustworthy read on private companies and the people inside them, so “with the lab models, I’d have to go buy all this data myself and point the model at that data lake, which would just be more work for me,” he explains. A model is only as good as the data behind it, and owning that data is a business of its own.

LinkedIn is broad, but it runs on what people write about themselves, and it thins out at exactly the small, early companies where BTV’s operators tend to sit. Harmonic is the layer built for this world: its coverage is startup-native and rebuilt continuously, and it resolves every company and person into one current record, so the answer is ready the moment the question is asked. “Harmonic is tailored to tech, which is the ecosystem we care about,” Nihar says. “It’s dense exactly where we operate, on the early companies and operators a general tool barely registers.” And it reaches BTV inside the same model the team already works in, through the MCP, so the firm gets the plain-language interface and the data beneath it without standing up a data team of its own. For a firm whose entire world is vertical AI and fintech operators, a dataset aimed at that world beats a broad index that treats a two-person startup and a Fortune 500 the same (if it registers the startup at all).

‍

The network is the product

BTV’s real product is its network. For a fund this concentrated, the relationships it can build and put to work are its edge. Harmonic is how a five-person firm creates a network it doesn’t have yet and then activates it, at a scale it could never manage by hand. The deals most funds work hardest to find come to BTV through that same network, as a byproduct of everything else. For a firm of former operators who owe tomorrow’s founders more than a check, the network is how they make good.