David Tisch on Seed Investing, Ownership and Finding Outlier Companies

Harry Stebbings welcomes David Tisch of BoxGroup back to the show after two and a half years. They return to a disagreement about Tisch's model: smaller ownership stakes and a broad portfolio. Stebbings says that he had criticized the approach and now believes he was wrong.

Asked whether venture investing has fundamentally changed, Tisch says both yes and no. He has invested at the early stage for 17 years, and believes investors too readily compare a new market to a previous one. To him, the present is a distinct set of conditions, not simply a repeat of 2021.

The underlying work, however, remains familiar: identify and meet founders, get to know them, decide whether to invest, and persuade them to accept the money. Investors have to perform that work within the conditions they face rather than expecting the market to conform to a past cycle.

Stebbings contrasts the speculative 2021 market with the current wave of technological change. Tisch resists choosing between 'bubble' and 'early opportunity': every period contains good and bad companies, and the important question for an early investor is whether a specific founder can build an exceptional business.

Tisch views macro conditions as useful for analyzing an industry or portfolio, but insufficient for judging a single company. Some highly valued startups will fail spectacularly; others will disappear quietly. At the same time, he believes many strong teams are building for the long term.

He acknowledges that companies are raising faster, at higher valuations and with less dilution. For him, those conditions are not automatically harmful across the industry. The decisive question is whether founders can navigate complex, rapidly changing conditions and create lasting value.

Tisch describes AI as a horizontal technology undergoing repeated waves of progress and reaction. What feels different is the speed: model releases, revenue milestones and acquisitions can create mini-cycles within a few months, making it harder to project the future from a static snapshot.

Stebbings asks how seed investors can adapt when promising teams routinely raise $10 million to $20 million, sometimes much more, before there is much operating history. A large entry check can consume a meaningful share of a small fund, changing the apparent risk of an individual investment.

Tisch says investors cannot set the terms of the whole competitive market. If another investor will fund a founder at a particular price, the choice is whether the opportunity is worth accepting at that price. Complaining does not remove the need to make that yes-or-no decision.

He offers two reasons larger entry rounds can still work economically. First, the potential outcomes for some companies have expanded dramatically. A valuation once treated as an extraordinary outcome may now be part of an investor's working scenario for an ambitious early-stage company.

Second, Tisch says some rapidly growing companies dilute their early shareholders less in subsequent financing rounds. He gives an illustrative range of 3–8% dilution in later rounds, compared with 12–20% in the past. Across several rounds, retaining more of the original stake can materially affect returns.

That does not mean every expensive deal is attractive. Tisch says a portfolio made entirely of very expensive seed rounds would change the firm's economics. BoxGroup looks for balance across different entry points even when it decides that one unusually priced company is worth backing.

Stebbings accepts that he underestimated how much potential outcomes could expand. In his earlier criticism, he did not fully account for the returns a relatively small check could generate if a company became vastly more valuable than expected.

Tisch describes venture as a power-law business: a small number of outliers matter disproportionately. Requiring 10–20% of a true outlier at seed means finding it, winning against other skilled investors and persuading its founders to accept a large stake. He considers that a demanding chain of assumptions.

He says established multistage firms have become more accessible, efficient and focused on seed than in the past. Their brands and long-standing relationships attract founders. In his view, a small firm should be cautious about designing a strategy that depends on consistently beating those firms head-to-head.

Fundraising has also become more efficient since the pandemic, Tisch argues. Deals that once attracted three to five early term sheets may now attract many more, or a deliberately limited group of highly competitive offers. BoxGroup therefore tries not to let rigid fund requirements prevent it from backing founders it considers exceptional.

Tisch tells Stebbings that critics miss both the fund mathematics and the competitive selection problem. If early ownership is diluted less in the strongest companies, a smaller initial position may retain more value than older models assume. Still, he notes, even great companies can experience setbacks or down rounds.

He challenges concentrated managers' claimed win rates. In his view, a fund that ends up owning 10% of each of 30 companies may have had to pursue far more opportunities to obtain them. The resulting portfolio is not necessarily the 30 companies it would have chosen if it could win every deal.

Stebbings worries that investors can distort companies by offering a second financing tranche at a much higher price soon after the first, leaving founders with too much capital and distraction. Tisch agrees that misuse is possible but rejects applying the same fear to every founder.

For some founders, Tisch argues, more runway and momentum can help with recruiting and customer partnerships. Founders who accept large amounts of capital and high valuations must execute differently from those who are constrained. The outcome depends on how they use those conditions, not merely the money raised.

Tisch criticizes the lean-startup playbook as a universal prescription for venture-backed companies. Stebbings points to bootstrapped counterexamples. Tisch distinguishes those businesses from a venture-scale trajectory, arguing that successful venture-backed companies eventually need significant capital; he says well-funded does not have to mean undisciplined.

The pair debate sequential financing tranches. Tisch would prefer to invest at the earlier, lower price, but says an investor must earn that allocation from the founder. He separates his emotional dislike of a later, more expensive entry from the business question of whether to invest anyway.

Tisch also distinguishes a founder's decision from a competing investor's deal structure. If another investor repeatedly pushes an existing backer into a less favorable tranche, he says, that can affect their future relationship. Venture firms depend on one another over time, and he believes short-term tactics have long-term consequences.

Tisch says BoxGroup has tried to remain consistent about early-stage investing and being useful to founders. It does not demand a fixed ownership percentage, does not routinely take board seats and can lead or participate in rounds. The aim is to avoid putting the firm's own constraints ahead of the founder's needs.

Initial ownership varies widely. Tisch says some investments begin at 1%, others at 10%, and many fall in between. He may be disappointed to write a very small check, but he treats the decision as an economic choice rather than a rule that automatically excludes the opportunity.

A particularly small allocation often means the firm arrived too late, after a financing was already underway. Tisch sees that as a sourcing problem: the team should ask how it could have met the founder before the company or financing process had formed.

Asked about BoxGroup's win rate, Tisch estimates it varies between roughly 70% and 90% with market conditions. He says arriving too late is the biggest controllable cause of losses, followed by the speed with which the firm can make a decision.

Some founders simply do not choose BoxGroup, perhaps because they do not connect with the team or believe another investor is more useful. Those cases are harder to correct. The firm instead focuses on meeting promising people earlier and improving access to opportunities.

Tisch describes a team of ten investors, including six partners who have worked together for years. Stebbings says he too has become more flexible after passing on strong businesses because they did not offer enough ownership. The ability to participate without leading can make an investor more competitive.

Stebbings asks whether lead-only seed firms still have a viable business. Tisch does not want to follow that model. He believes it is difficult to compete for a large stake against excellent multistage firms that can also lead seed rounds and have established founder relationships.

Stebbings adds that some larger firms can offer a much higher price and a bigger check. Tisch emphasizes that these are offers in a competitive market. The founder is entitled to accept one, and an investor who insists on a particular stake must reckon with the possibility of losing the deal.

The willingness to wait for a later round has also changed. Tisch says a firm that passes at seed may never get another chance: competitors may fund the next round, too. A firm that loses an early contest might therefore move quickly into the next financing even if the company has made limited new progress.

Stebbings says many limited partners seek small, concentrated, pure-play seed funds, even though he considers that a particularly hard segment. Tisch describes three groups with different incentives: founders have leverage, general partners face a changing market, and LPs may want exposures that are difficult to obtain cleanly.

Large multistage firms can absorb substantial LP capital and offer access to established venture franchises, even when LPs express reservations about their blended products. Tisch says there is no simple solution to these competing needs or a perfectly static strategy for general partners in a changing market.

BoxGroup's current seed fund is $275 million and is intended to invest in roughly 150 companies. Tisch says some LPs do not believe the mathematics of a broad portfolio and smaller ownership stakes can work. He argues that the firm's own results and its existing partners support the model.

Asked about realized returns, Tisch points to Cursor. A BoxGroup partner knew a founder from their work at Dorm Room Fund, and the firm invested in an early round. Tisch confirms that the investment returned the fund multiple times, while describing the firm as fortunate to have met the team early.

Stebbings notes how quickly some startups reach large valuations. Tisch agrees that company creation and growth can be unusually fast, but cautions against treating Cursor's path as the only valid timeline. Other interesting businesses in the same fund are compounding on a slower schedule.

He cites Clay as a company that went through a bumpy period before explosive growth, learning and iterating on go-to-market along the way. He also mentions Baseten, which needed inference to develop as a market before its own growth accelerated. Neither was simply an uninterrupted line upward.

Asked whether the best companies are obvious immediately, Tisch looks for evidence of exceptional execution: recruiting outstanding people early, reaching customers who seem hard to access, or combining a strong product with effective market conversations. Those signs do not eliminate the challenges ahead.

Founders must navigate uncertainty rather than merely predict a stable future. Tisch recalls asking students to describe the world in ten, five and two years; fewer were willing to make longer-range predictions. Rapid model releases and sharply different visions of AI's future make the environment especially difficult to forecast.

AI adoption will reach industries at different speeds, while computing, energy, data centers and politics create additional uncertainty. Tisch says an early-stage investor is looking for founders who can respond to complicated conditions in real time, not for someone with a guaranteed prediction.

Stebbings asks whether the market might later value sales and go-to-market skills more highly again. Tisch notes that an early-stage bet takes years to mature; an investor must make decisions about the future market rather than expect an immediate result from today's opportunity.

Tisch believes many leading founders today are more technical and capable than before. In his view, there is less room for mediocre execution. Starting a business remains possible for many people, but a venture-backed company must pursue outcomes large enough to support the economics of venture funds.

Asked about consumer brands, Tisch praises the founders of Harry's but says BoxGroup has stopped funding brands at seed. He wants evidence that a product and brand resonate with customers, including purchase and acquisition economics, and believes that information is usually unavailable at the earliest stage.

He does not dismiss brands as small businesses. He points to beverage, quick-service restaurant and personal-care companies that can scale meaningfully, and suggests that a specialized investor entering at Series A may be better positioned to evaluate their initial traction.

The hosts discuss how capital can concentrate on fashionable categories. Tisch counters that founder quality and execution still matter, using Rogo, an AI-for-finance company, as an example. After BoxGroup's seed investment, he helped the company approach investors for a Series A.

According to Tisch, 39 of the 40 investors approached passed, while an investor named Keith led the round. The company later attracted other prominent firms. Tisch says the experience was frustrating but illustrates that early-stage markets do not have perfect information and a company can be misunderstood before building momentum.

BoxGroup now operates a $275 million seed fund and a $275 million follow-on fund. Tisch says the latter makes roughly 20 to 30 larger follow-on investments, mostly in existing portfolio companies, with checks of about $5 million to $30 million.

Sometimes the firm cannot increase a position because later allocations are limited or a founder chooses another investor. In other cases, Tisch acknowledges, BoxGroup misread a company or reacted too slowly. He regards continuing to help the founder as the central responsibility regardless of the size of a subsequent check.

On pro rata rights, Tisch rejects both treating the provision as irrelevant and assuming it guarantees an investor an allocation. He believes the right to participate in future rounds is ultimately earned through the relationship with the founder, whether or not it appears in a contract.

A founder who asks an early investor to step back creates a difficult choice. Tisch says insisting on a contractual right can carry financial and reputational consequences. Conversely, documenting the right can give a founder leverage when negotiating with a new investor who wants a large share of the round.

Tisch observes that firms may claim to require a certain ownership percentage in one deal while accepting less in another. In his view, 'need' often means a preference rather than an absolute constraint. Individual investors within a large firm can also behave differently depending on their experience and incentives.

Relationships among investors matter as well. Tisch describes cases in which a firm receiving an introduction later tries to squeeze out the introducer. He distinguishes that kind of short-term behavior from the durable relationships that help investors work together across many companies.

Asked whose recommendations receive the most attention, Tisch says founders with good judgment and strong networks are especially valuable sources. He does not limit that to the most visibly successful founders: a trusted assessment from someone who knows a prospective founder can be a strong starting point.

Stebbings turns to secondary sales and returning money to limited partners. Tisch says he works for the LPs who supply capital while founders are the firm's customers. Selling shares can produce returns, but it also creates financial risk and can affect the relationship with a founder.

Generally, BoxGroup does not want to sell early. An investment may last four, 15 or 20 years, and Tisch believes the best companies can compound substantially later in their lives. Selling too soon may sacrifice future returns as well as damage trust if a founder objects.

He does not assume that a founder taking some liquidity gives an investor an equivalent reason to sell. A founder may have debt, stressful living conditions or a need for a financial safety net. A modest personal sale can help that person continue building with greater confidence.

Tisch also sees periodic employee tender offers as potentially helpful for retention and culture. Employees may stay engaged when they know there is a path to some liquidity. He treats those personal circumstances as distinct from a fund selling shares to improve its reported results.

Asked what he has changed his mind about, Tisch jokes that every firm presents itself as a top performer, even though industry returns are concentrated among relatively few. He wonders where all the firms outside those top groups are, given how rarely they describe themselves that way.

On whether there will be more venture firms, Tisch suggests the market has less new-firm noise and more competition than it appears. A firm can remain visible by name long after its investing activity has slowed or its ability to raise another fund has diminished.

He returns to the idea of a small seed fund. Many investors begin with a modest first fund but hope eventually to manage far more capital. As a firm scales, its check size, stage, concentration and investment style may all change, even if its initial pitch sounded narrowly focused.

For that reason, Tisch asks new managers what they want their fourth fund to look like. He considers that a better expression of their long-term ambition than the strategy forced by the size of their first fund. Advice should help them reach the business they actually intend to build.

He says BoxGroup has increased its scale but kept an early-stage identity: finding founders early and writing initial checks. Stebbings notes that the two current funds total $550 million. Tisch acknowledges the scale while emphasizing what he sees as continuity in the firm's investment approach.

Stebbings cautions that building a much larger organization does not necessarily make its managers happier. Tisch replies that scaled firms have different structures and partnership cultures. He names several newer and established firms he admires and resists treating all multistage models as identical.

To Tisch, newer firms that combine ambition with strong execution can still create a durable position. Others will fall away. The relevant question is how an individual organization builds its team, relationships and investment practice, not simply whether it fits a broad label.

Stebbings asks Tisch to rank finding, picking and winning investments. Tisch says the firm's daily obsession is seeing opportunities: getting in front of founders and being taken seriously. He estimates the team sees perhaps 50–75% of what it would ideally want to see, while acknowledging the uncertainty in that estimate.

Winning a deal is an area where he thinks investors can improve tactically. The team asks what a founder wants and which relationship-building approach fits. Some founders seek transactional help, others want advice or a close confidant; Tisch views venture investing in part as a service business.

Picking is more mysterious. Tisch disputes the assumption that a broad portfolio means BoxGroup says yes indiscriminately. The team tries to recognize the rare opportunities it should back, but he finds it difficult to specify a repeatable method for improving that judgment beyond experience and developed taste.

Tisch says the firm meets many companies and gathers as a team twice a week, while retaining flexibility to make decisions outside meetings. He dislikes groupthink and does not use a vote as the sole basis for a deal: someone on the team must have genuine conviction.

He describes an internal rule against throwing 'grenades' at early-stage deals. A colleague who has not spent time with the founders should not casually dismiss another partner's opportunity. Concrete information about ethical or reputational concerns, however, is important to share.

His reasoning is that it is easy to list plausible reasons almost any young company might fail. Tisch wants partners to identify the conditions under which a company could work and decide whether they believe in the founders' view. The job, he says, is to say yes to the right opportunities.

Stebbings raises the question of how frequently early investments fail. Tisch believes seed-stage losses can be substantial and that risk falls once companies reach later stages. He also notes that an unusually talented investor may have a lower loss rate than the broader market.

In a rapid-fire discussion about venture funds, Tisch highlights Y Combinator's recognizable product, brand and ability to attract founders. Asked to name a single Series A firm, he instead lists several strong organizations and says fund sizes, stages and LP exposure make a simple ranking difficult.

He names a range of growth and multistage investors rather than choosing one definitive winner. Stebbings presses him for a single answer and offers his own preferences. Tisch's broader argument is that several different organizations have built effective but distinct models.

Tisch expects leading venture firms to continue compounding advantages in access to founders, capital and returns. He anticipates an uneven industry in which a limited group of firms is highly competitive and an even smaller set stands out within that group.

Asked about memorable founder meetings, Tisch mentions a colleague's early meeting with an Instinct founder. He says he tries to approach each investment with a consistent threshold of excitement rather than sorting every new founder against a single favorite.

He describes the variety of founders he meets: people attempting a difficult game, a household hardware product, scientific work or a foundational model for a physical-world application. Different businesses face very different risks, but each investment must be compelling enough to earn the firm's conviction.

Tisch says he particularly enjoys conversations with consumer founders because of the challenge of understanding end users. He mentions Triumph as a company whose founder and work he finds interesting, while acknowledging that such early-stage outcomes remain uncertain.

Looking ahead, Tisch is excited by the uncertainty itself and by meeting people who may shape the next decade. He cites autonomous driving as an area where he believes adoption could bring significant safety benefits. Stebbings pushes back that challenging cities and weather remain difficult; their exchange presents differing judgments about readiness.

Both discuss the tension between building new institutions and improving existing ones. Tisch says society should pursue practical decisions and hopes that technological abundance will benefit a broad population, including through advances in health and science. They close by revisiting their good-natured dispute over ownership and diversification.