Avlok Kohli on Venture Funding, AngelList and Building for Speed

Avlok Kohli, CEO of AngelList, tells Harry Stebbings that a company is a group of people building products for customers. His prescription is to remove meetings and processes that slow delivery, then iterate quickly with customer feedback.

Kohli says AngelList co-founder Naval Ravikant invested in each of his three startups. After the last was acquired, Kohli expected to focus on investing, but Ravikant asked him to consider becoming CEO. He initially declined; after six months of conversations he accepted in July 2019.

His delivery startup Fastbite promised 15-minute delivery and averaged seven minutes, Kohli says. Building for himself helped him understand customers, but the experience also showed that optimizing a single variable such as speed can miss the wider product experience: sometimes customers prefer a slower delivery.

Being first to market does not guarantee success, he argues. The first entrant may map an ‘idea maze’ for later founders without successfully navigating it; he points to Google and the iPhone as products that were not first in their categories.

Kohli defines defensibility as the cost of replicating a business over time. On day one, he says, there is effectively no business to defend, and even a quickly assembled network may be replicable. He is skeptical of treating early-stage defensibility as an established moat.

Asked about European rapid-delivery startups, Kohli warns that logistics and online-to-offline businesses require substantial capital. He predicts they will fail in a downturn because higher capital costs make it hard to finance expansion and the early losses preceding scale. This is his sweeping forecast in 2023, not a verified outcome for every company.

Kohli says AngelList supports 20,000 funds and syndicates with a portfolio of more than 13,000 startups, and receives investment documents as signatory on behalf of funds. That gives the company a particular view of venture activity, although its observations are based on its own platform and need not represent the whole market.

AngelList tracks the share of startups receiving investment and the share of those investments that are up rounds. Kohli describes early 2022 and 2021 as boom conditions, and the current quarter as a deep bear market in its data. He says the shift between the two was unusually large and fast relative to earlier observations.

Stebbings says seed prices still look high. Kohli distinguishes seed and pre-seed from later rounds: he says Series B deal volume and valuations were each down more than 50% in AngelList’s view, with Series A also down and late-stage deals rare. He attributes stronger seed pricing partly to later-stage investors moving earlier, and expects seed valuations to remain relatively resilient.

Stebbings suggests junior investors at multistage firms have also entered seed with less price sensitivity. On Series A, Kohli says companies with credible traction can still raise, while others cannot; AngelList’s data did not let him distinguish the quality of companies with insider-led rounds from those with new outside leads.

They define a party round as one in which no investor writes more than 40% of the round. Kohli says its share of rounds had stayed broadly stable even as total volume fell. Repeated small extensions had become much harder, but a round without a lead could still close, especially at smaller seed sizes; a $5 million round is harder to assemble this way than a $2 million one.

Fundraising and time to markup are taking longer, Kohli says. Comparing a sample of recent investments with historical averages since 2015, AngelList saw roughly 33% fewer companies than expected having raised by that point. He reads this as a freeze relative to history and expects repricing to take several more quarters.

Kohli says AngelList had seen 48 flat rounds, down rounds and pay-to-play transactions in the first quarter so far. He expects more as companies that extended runway through layoffs and cost cuts eventually approach cash-out dates. His projection for the following quarters depends on market conditions; the figure is a count, not a percentage.

Despite expectations of a consolidation spree, Kohli doubts acquisitions will rise rapidly. Public companies face pressure over cash generation and stock-based compensation, while private buyers and sellers must agree on valuations. Even a seemingly free acquisition brings integration and personnel costs; he says AngelList’s own data showed M&A had fallen substantially.

Kohli expects the next six months to remain difficult if interest rates stay higher for longer. Venture competes for capital with safer yields, he says, while recovery over an 18-month horizon would depend on rates, inflation and the broader economy. These are his forecasts at the time, not outcomes.

AngelList has seen more interest in selling fund positions, including stakes linked to well-known companies. Sellers may need liquidity or need to rebalance illiquid holdings. Kohli says some fund-position deals cleared at discounts starting around 20%, more commonly 40% to 50%, relative to the last-round mark; he is discussing fund positions rather than direct secondary sales.

For rolling funds, Kohli says commitments from both individuals and institutions were down around 20%. For traditional venture funds on AngelList, individual commitments were down around 60% from early 2022, while institutional commitments had fallen and recently recovered in its data. He uses ‘institution’ broadly to include family offices with teams.

Kohli hypothesizes that greater macroeconomic certainty and excitement around AI help explain the recent institutional recovery. He describes that explanation as a hypothesis, not a measured causal finding, and says large language models have already affected AngelList’s operations.

The number of new funds raising had fallen, he says, even as more experienced funds could attract more capital. Time from the first dollar to a first close had grown from roughly three months in boom times to over six months and was trending toward seven. Smaller and first-time managers were also raising smaller funds; a prospective $5 million fund might become $1 million to $3 million.

AngelList manages fund documentation, setup, filings, portfolios and distributions over long periods. Kohli says large language models allow the company to automate some tasks that previously required human judgment, drawing on years of internal data.

As an example, he estimates that AngelList receives about 15,000 emails a week on behalf of funds. Its system can examine email and legal-document contents, identify a share purchase agreement or cap table, route work and automatically request a missing document. He presents this as one example of its automation, not a claim that every decision is automated.

Stebbings asks why manager pricing increased if AI can lower costs. Kohli says AngelList had expanded from small funds into more complex $50 million, $100 million and $200 million funds. A pricing cap originally designed for much smaller funds was removed and pricing adjusted to fund size; he acknowledges the rollout may have caused confusion.

Kohli says fund administration is not a loss leader and automation has expanded margins. AngelList evaluates its economics across the entire platform, with products for general and limited partners feeding one another. He claims margins above 80% across the overall product experience; he does not give a separate margin for every product.

He credits lessons from Square, which acquired Fastbite: a payments product grew into a platform where products reinforced one another. Kohli says funds have AngelList’s best margins, whereas some smaller special-purpose vehicles are loss leaders.

He argues that administering small SPVs is harder than it looks because the provider must support complex instruments for years and unexpected costs may appear later. He cites the failure of SPV provider Assure as a cautionary example, in his own account, and advises managers to consider long-term reliability alongside price.

Asked which product he regrets, Kohli points to startup incorporation. AngelList later partnered with Stripe Atlas; in hindsight, he says, the company should have asked whether it could contribute a distinctive solution. He contrasts that with products where AngelList believes its combination of legal, product and founder expertise is unique.

Kohli says he scaled back and closed AngelList’s European operation to concentrate resources on the US, where it had not yet reached scale. European regulatory structures were less standardized, in his account. He wants the company to secure the US market and develop a clear standardized approach before returning to Europe.

He defines winning the US as supporting more than 50% of the market across venture-fund and venture-backed startup segments. He does not disclose company-side market share, but claims AngelList often reaches 50% to 60% quickly when it enters a particular fund-size segment. These are the CEO’s claims and targets, not independently verified market measurements.

Stebbings raises direct competition with Carta. Kohli says AngelList watches competitors but concentrates on creating a better customer experience when it has a distinctive view. He says it wins both newly formed companies and customers moving from other providers; he does not give a breakdown.

To execute quickly, Kohli emphasizes ambitious hires, clear deadlines and tight feedback loops within working teams. Remove blockers, he says, cancel unnecessary meetings and ship iterations to customers. His earlier opening remarks make the same case for speed over process for its own sake.

Kohli says he does not underestimate small, ambitious startups as competitors. Stebbings asks whether a shrinking venture market is the larger threat; Kohli responds that AngelList was then stronger than ever, citing movement into larger funds, customers switching from other providers and diversified products. This is a company leader’s assessment of his own business.

In the quick-fire round, Kohli predicts conditions will be worse by the end of 2023 but better in 2024. He expects AI to amplify knowledge work and require extensive retraining. He favors startups as potential beneficiaries because of founder-led product craft, while recognizing incumbents’ distribution advantage.

Asked which investors he would buy, he names Sequoia among multistage firms and Haystack among seed specialists, declining to name firms to short. He also says he no longer believes a great team can reliably find a great market on its own: a strong team and a suitable market must come together, and a pivot requires enough capital.

Kohli sees insufficient diligence and fear-of-missing-out investing as major mistakes of the prior 18 months. He would like limited partners to look beyond established names and assess other signals; he says AngelList’s data include strong first-time fund managers and discloses that he invests as an LP in first-time managers himself.

Asked about the next US election, Kohli says he does not expect Donald Trump to win and hopes for a different candidate from Joe Biden. This was his 2023 political prediction and preference. He names Gokul as a favorite angel investor for experience spanning early startups and enduring companies.

Looking ten years ahead, Kohli wants AngelList to become infrastructure for venture funds and many venture-backed startups, moving ownership and investment activity from spreadsheets into software. Asked whether the company will go public, he says he does not know; his focus is building an enduring company with compounding cash flows.