Dan Siroker on Repeat Founders, Fundraising and Building for Real Problems

Dan Siroker tells Harry Stebbings that founders often mistake fundraising for a simple question of whether an investor will say yes. They can also negotiate price and terms, provided they understand what the investor needs and the leverage available to both sides.

Siroker traces his early interest in computers to a mentor, Hector Garcia-Molina, his mother's boss. He describes a recurring desire to do things others consider impossible, a motivation that extends beyond business to his personal life.

A startup can feel like climbing a mountain without knowing whether it is the right mountain. Siroker recalls Elad Gil's observation that promising ideas often reveal early glimmers of traction; Dalton Caldwell described a good pivot as feeling like coming home. Stebbings compares that with his own experience of trying ideas until one feels compelling.

Having founded Optimizely before building Limitless changed how Siroker thinks about company building. First-time founders can confuse a growing headcount with progress, he says; a small team focused on a few consequential problems can often accomplish more.

In his experience, founders must retain their own judgment while absorbing advice. Nonconformity is valuable, but refusing to learn from experienced people can create avoidable mistakes. The challenge is to listen carefully without letting someone else make the decision for you.

Siroker distinguishes accountability from abdication: a founder can give someone responsibility for a task while remaining accountable for the outcome. He believes leadership requires identifying the handful of things that matter most and staying close enough to know whether they are working.

He generally favors repeat founders as an investor because they have experienced setbacks, built networks and demonstrated perseverance. He acknowledges that exceptional first-time founders exist; experience is a preference and a source of information, not a guarantee of success.

At Optimizely, Siroker says he sometimes followed other people's advice against his own instincts. In hindsight, pushing toward enterprise sales too early and failing to pay enough attention to customer churn were expensive lessons.

Optimizely's self-serve product could bring in a large company through an individual user: he recalls a Starbucks customer paying $79 a month. That kind of product-led adoption may reveal demand before an organization builds an expensive enterprise sales operation around it.

Stebbings asks about job titles. Siroker prefers flexible 'head of' titles early on, when the future shape of a company is uncertain, rather than immediately promising executive titles that may be difficult to sustain as the business grows.

He regards excessive concern about title as a potential warning sign but also recognizes that titles have real value to founders and employees. Compensation should be fair: he describes raising teammates' cash pay by $30,000 when he learned a new hire's salary would otherwise exceed theirs.

During an oversubscribed Limitless Series A, Siroker says employees could sell up to 25% of their vested shares. He sees voluntary secondary sales as a way to give people financial breathing room and potentially improve retention, while recognizing that a buyer and an appropriate transaction are required.

His hiring preference is for people who have already acquired relevant experience rather than employees straight out of college. He compares a roughly 20-person team at Limitless with the much larger Optimizely organization and argues that a small, experienced group suits the company he wants to build. This is his approach, not a universal rule.

A venture investor's conventional view is that a lead needs about 10% ownership. Siroker argues that a founder with a strong position can sometimes negotiate a smaller allocation. The investor's incentives matter, but they are not necessarily a fixed law.

He recalls Benchmark taking roughly 8% in an Optimizely round in 2013, with a lower valuation as part of the negotiation. His lesson is to understand the other side's economics and use the available terms creatively, rather than assuming ownership, price and the investor's contribution are unrelated.

Boards can provide particularly valuable guidance to first-time founders, Siroker says, but the composition and usefulness of a board depend on the people involved. He encourages founders to distinguish an investor's reputation from the practical help and judgment they will bring.

Siroker describes publishing a fundraising pitch and drawing several million views. He says the process produced substantial interest, including 22 offers at a $1 billion valuation; these are his account of the process, not independently verified completed transactions.

He says he ultimately selected NEA at a $350 million valuation and later gave other interested investors a way to participate through a pooled vehicle. His point is that the highest headline valuation is not automatically the best choice: who leads, how interests align and what happens in future rounds matter too.

Siroker argues that an investor's question about how much a company intends to raise often implies an ownership target and therefore a valuation. He favors transparency about those assumptions, while acknowledging that negotiations and the parties' incentives can constrain how open a conversation becomes.

To make fundraising more efficient, he recommends concentrating first investor meetings into a defined week. Practice with investors who are less likely to lead, listen to their questions, improve the presentation and keep supporting detail in an appendix.

Early-career investors can be particularly motivated to help a company succeed, in Siroker's view. He has experienced both helpful and less helpful investor support; ultimately, he says, the founder still owns the company's outcome. A recognizable investor brand can nevertheless help with recruiting and market perception.

Founders should investigate investors through references, including people whose businesses did not become obvious successes. Siroker values candor and practical help over the promise of a famous name; Stebbings suggests that even a brief response to an investor update shows a founder that someone is listening.

Siroker warns against treating outcomes as only zero or a multibillion-dollar success. At a sale for around two or three times the capital raised, liquidation preferences and the size of the financing can produce very different results for preferred investors and employees holding common stock.

He cautions against hiding a difficult financing behind an apparently higher valuation paired with a multiple liquidation preference. His advice is to consider the distribution of proceeds and, when faced with that trade-off, favor a down round over adding an onerous preference. The right outcome depends on the actual deal terms.

Control is another negotiating issue. Siroker recommends that founders consider super-voting stock, board composition and observer seats, and advocate for common shareholders as well as themselves. He frames these as choices to negotiate, not as substitutes for a trustworthy relationship with investors.

Stebbings pitches his own investing proposition: he will avoid getting in the way, make introductions and support founders with additional capital when needed. Siroker likes the offer, and both agree that a short, thoughtful response to a founder update can make support tangible.

Siroker's most memorable venture meeting was with Benchmark's Peter Fenton in 2013. Fenton asked what would keep him excited about the business five years later. Siroker says he did not understand the significance at the time, but around 2018 he felt disengaged from Optimizely.

With hindsight, he interprets the question as a test of whether a founder can stay committed for a decade or more. He says the problems Limitless is tackling feel like his life's work, even if the company were ultimately to fail.

For investing in AI, Siroker looks for founders obsessed with a real problem rather than a particular technology. He sees 'wouldn't it be cool if' products as a familiar trap: the existence of a powerful tool does not by itself establish a customer need.

Limitless is building a wearable, but Siroker says he would gladly give up hardware if software could solve the same problem. His proposed filter is whether a founder understands a real need, has the determination to address it and can use the available technology, not whether the company belongs to a fashionable layer of the market.

A small or obscure problem can be attractive if a founder knows it firsthand. Siroker imagines a customer-service founder who has personally answered support calls: that experience may provide an understanding a market map alone cannot. He concedes that his preference could cause him to miss good companies.

Optimizely grew out of a problem he encountered working on the 2008 Obama campaign. Existing A/B testing tools had pain points, he says; he wanted a visual product that did not require a developer. Firsthand experience helped him decide which features were necessary and which could be left out.

Siroker believes a CEO must protect product simplicity. Engineers responding to individual customer requests can create useful features one at a time, yet a hundred such decisions can leave a product unfocused. Each feature also creates an ongoing maintenance commitment and edge cases, not merely an initial build.

He regrets missing adjacent opportunities in product analytics and customer data. He recalls trying unsuccessfully to buy Amplitude when it had four people and exploring an acquisition of Segment; in hindsight, he believes Optimizely could have pursued those markets itself instead of treating competitors' early leads as insurmountable.

Asked about launching Limitless's pendant, he says a launch should make the underlying problem unmistakable. A polished announcement can fail if people leave unable to explain how the product improves their lives.

Siroker says he probably should have returned to Y Combinator for his second company: pride may have made him underestimate the value of its network and resources. Asked whom he would add to a board, he names Sam Altman and Elad Gil for their judgment and willingness to offer candid advice.

On fatherhood, he advises his younger self to expect the early years to be demanding and to make room by cutting other commitments. Raising children while running a startup can be physically and emotionally exhausting, he says, but concentrating on his family and work has also made him happier.

With three children, Siroker treasures ordinary time together, from playing at home to planting things. He also enjoys watching children learn and thinking about the parallels and differences between human learning and the AI systems he works on.

Looking to the future, Siroker worries that pessimism about technology and regulation shaped by incumbent interests could slow innovation. These are his concerns and predictions, expressed in the 2024 conversation, rather than established outcomes; Stebbings questions whether AI development can be stopped so easily.

He notes that building foundation models requires substantial capital, which may make open competition difficult. Asked whether TikTok will be banned, he says he is not an expert and expects regulatory headwinds, while treating political speculation as uncertain.

For Limitless in 2034, Siroker imagines millions of daily users and a shift in attitudes toward capturing what people say and hear to augment memory. That is his company's aspiration, not a forecast of adoption. He wants the technology to support human intelligence rather than replace it.

Siroker hopes that, if the product succeeds, tools for remembering conversations will eventually feel ordinary. Stebbings thanks him for a candid discussion, and Siroker closes by thanking the host.