Venture capital has spent much of the past decade becoming a department store. One firm might offer seed checks, growth rounds, crypto tokens, recruiting departments, executive briefings and a podcast studio down the hall. Chemistry opened in San Francisco in 2024 with the opposite proposition. Its founders raised $350 million, enough to matter at Series A, and then promised to behave like a small partnership: make a select number of investments, bring all three managing partners to each one and do the unglamorous work after the wire lands.
The three partners arrived with unusually portable résumés. Ethan Kurzweil spent 16 years at Bessemer Venture Partners, investing in developer platforms, data infrastructure and companies such as Twilio and LaunchDarkly. Kristina Shen led B2B software investing at Andreessen Horowitz after seven years at Bessemer. Mark Goldberg spent nearly a decade backing software and fintech at Index Ventures, after an operating stint during Dropbox’s high-growth years. Between them, they had led nearly 100 investments, served on more than 50 boards and backed more than a dozen companies that became unicorns.
That history is the raw material. The actual product is access to it.
The constraint is the feature
Chemistry leads investments from the first check through Series B, with the center of gravity at Seed and Series A. Its customer is a technical founder who has more than a clever prototype: someone with the instinct to turn technical leverage into a business buyer, a distribution advantage or a new operating model. The fund’s public language repeats that pairing - technical expertise plus commercial instincts - because it is broad enough to cross sectors but narrow enough to reject science projects without a market.
For founders, the problem Chemistry is trying to solve is familiar. A prestigious partner can be deeply present during a financing, then become one name among dozens in an expanding portfolio. Larger platforms have genuine advantages - specialist recruiting, marketing benches, global offices and capital across stages - but attention is not infinitely divisible. Chemistry’s answer is to limit the denominator. The firm said at launch that each partner would back roughly two or three companies a year, and that all three founders would support every investment.
“We are the portfolio services team, working in the trenches with our founders.”Chemistry’s launch statement
This is also a business-model choice. Chemistry manages money committed by limited partners, buys minority stakes in private companies and aims to return more than it invested when those companies are acquired, sell shares or go public. Like other venture firms, it likely earns management fees and a share of profits, though its exact terms are private. What is unusual is not the economics. It is the promise to resist the throughput that those economics can encourage.
Institutional memory, compressed into one small loop
selection
work
Series B
A portfolio of stubborn bottlenecks
Chemistry’s portfolio makes the thesis easier to see. Assort Health uses voice AI to handle the phone calls that clog medical offices. When Chemistry co-led its 2025 financing with First Round Capital, the firm said Assort had cut hold times by more than half, reached resolution rates above 90 percent and helped hundreds of thousands of patients. The buyer is not purchasing “AI” in the abstract. A clinic is buying fewer abandoned calls, more accurate scheduling and staff capacity.
Datacurve supplies complex coding data to foundation-model labs. Nova Intelligence builds AI agents for SAP, where proprietary code and deeply customized enterprise environments frustrate general-purpose coding tools. Yuzu Health rebuilds the administrative core of health plans - claims, member administration and provider payments - rather than adding a pretty interface to old infrastructure. Didero attacks procurement. Serval rethinks IT service management. Noon merges design and software creation on a dual canvas.
The categories stretch from enterprise software to healthcare and, most recently, physical machines. In July 2026, Chemistry invested in Atoms, a company pursuing purpose-built autonomous machines for food, transport and mining. The round totaled $1.7 billion and was led by Andreessen Horowitz; Chemistry called its participation the largest check in its own history. That sounds far from an early-stage help desk company, but the filter is consistent: hard technical work attached to an enormous, legible operating problem.
A directional map of publicly described investment themes
Conviction, with a consumer streak
The partners’ writing provides a second map. Kurzweil argues that AI developer tools are escaping the developer market and acquiring consumer dynamics: natural-language interfaces, viral adoption and individual users who later pull corporate budgets behind them. Shen has tracked successive waves of voice AI, particularly where latency and reliability finally beat the call-center baseline. Goldberg has written about marketers losing search traffic to answer engines. The team has explored synthetic research, autonomous agents, AI security and the collapse of boundaries between design and engineering.
Those ideas explain why the firm fits awkwardly into a conventional sector box. It is an enterprise and developer-tools investor by lineage, a fintech investor through Goldberg’s experience, and increasingly an AI investor by market reality. Its healthcare bets are often software and infrastructure bets in disguise. Its robotics investment is a wager on autonomy becoming operational. The common expertise is recognizing when a new technical capability has crossed the line from demo to dependable workflow.
One investment makes the culture unusually tangible. Granola, the AI meeting notebook, became central to Chemistry’s own workflow shortly after the firm launched. At a May 2025 offsite, the small team reportedly voted it the company they most wished they had backed. They sent the founders a 19-point product wish list. Months later, Chemistry invested. The anecdote is funny because it reverses the normal diligence theater: the investor did not have to imagine customer love; it was the customer, with an embarrassingly detailed feature request.
The market position - and the test
Chemistry competes with the institutions its founders left, along with focused early-stage firms such as First Round, Conviction, Uncork and Afore. It also co-invests with some of them. The competitive line is therefore less about exclusive sectors than service level. A founder can choose the enormous platform and its specialist bench, or choose a smaller partnership that claims fewer internal handoffs and more senior time. Chemistry has enough capital to lead meaningful rounds, while keeping the team page short enough to read over coffee.
That makes Chemistry most useful at a particular moment. A founder who already knows the category and needs a recognizable lead can tap a partnership with experience in product positioning, executive hiring, enterprise introductions and the emotional weather of a boardroom. The firm is not an accelerator, a development shop or a source of non-dilutive financing. It buys equity and expects venture-scale outcomes. Its support can improve the odds, but it cannot replace customer demand, sound unit economics or a team that ships. The cleanest way for a founder to evaluate the offer is to reference-check the individual partner, ask how decisions are shared and establish exactly who will do what after the round.
The fund’s size gives it room to reserve capital for later rounds without having to become a growth investor. It also raises the bar for ownership and outcomes: a $350 million vehicle needs meaningful winners. Chemistry has responded by leading rounds rather than collecting passive logos. Datacurve’s $15 million Series A, Assort Health’s financing above $22 million, Yuzu Health’s $35 million Series A and Nova Intelligence’s Series A all put the firm in a visible price-setting role. Other portfolio entries are labeled “pre-Chemistry,” an important distinction between the partners’ inherited track records and the young firm’s own scorecard.
Its culture mixes institutional vocabulary with startup rituals. The firm runs Elements, a San Francisco conversation series whose guests have included OpenAI COO Brad Lightcap and Decagon founder Ashwin Sreenivas. Its public feed shows reading groups, office gatherings and even a NeurIPS surf camp. The name itself came from a word the three founders kept using about their relationship with one another and with entrepreneurs. Planning moved from nights and weekends into an always-on WhatsApp thread. The brand arrived after the behavior.
The unresolved question is scale. A promise built on three people’s direct attention works precisely because those people are scarce. A second fund, more partners or more follow-on obligations could strain it. Chemistry’s founders acknowledged at launch that growing the partnership might recreate the complexity they left behind. This is not a flaw hidden in the model. It is the model’s central tension.
For now, the evidence is a young portfolio, a run of visible Series A leads and a willingness to publish what the team believes before outcomes are known. Founders can use Chemistry for capital, board-level judgment, customer and candidate introductions and a working relationship with investors who have watched companies move from first check to public market. Limited partners are buying the founders’ accumulated pattern recognition. Chemistry is betting both groups would rather have a concentrated dose.