Julien Bek | 20VC Summary

1. Key Themes

Sequoia Is a Hunting Firm, Not an Inbound Platform

The perception that Sequoia waits for elite founders to call is flatly wrong. Every partner is expected to source and win deals independently, regardless of tenure.

"Everyone thinks that we're just waiting for the phone to ring for the next Anthropic to call us to invest. That's completely false. Everyone at Sequoia is a hunter." 00:00:00

The firm operates with roughly 11 people on the early team — the size of a football squad — and everyone is expected to score on the field.

Conviction Over Consensus: The SpaceX Lesson

The best investments in Sequoia's history are consistently the ones where the sponsor had the highest individual conviction, not the ones that passed IC smoothly.

"The best investments in all the funds are always the companies where the sponsor had the highest conviction. That's just the one thing that's been true time and time again across funds." 00:00:00

Sean McGuire's SpaceX investment received a vote of "one" — the lowest possible score — from at least one partner. McGuire refused to give up, forced the entire partnership to fly out and see it firsthand, and Sequoia made a smaller initial investment that grew into one of the greatest in the firm's history.

Updating Priors Is a Competitive Skill

The ability to revisit a prior rejection and re-underwrite at a dramatically higher valuation — overcoming ego — is treated as a core discipline at Sequoia, exemplified by the $2.5 billion Anthropic check after an earlier pass.

"It came from the realization that if AI is going to be so transformative, we are just on the foothill of this incredible exponential... the human brain is just not very good at dealing with exponentials." 00:10:16

Agents Are the New Customer — and Require a Parallel Economy

Julien's central thesis: agent traffic is already at parity with human traffic, and Cloudflare projects 1,000x agent-to-human traffic within five years. This creates a need to optimize for agents the same way companies spent 20 years optimizing UI for humans.

"We need to think of a bits-perfect platform that's good at converting agents." 00:48:35

Agents are not neutral — they already exhibit strong brand biases (toward Cloudflare, Vercel, etc.) baked in through pre- and post-training, and hedge funds are already buying data to understand agent decision-making patterns.

The Services-to-Software Transition Is the Next Trillion-Dollar Opportunity

Julien's viral thesis: the next trillion-dollar company will be a software company that masquerades as a service business — capturing the $6 of service spend alongside the $1 of software spend, but doing it with software-like margins over time.

"Today you might buy QuickBooks for $2K, but you spend $15K to close the books with your accountant. So what if you can just sell the outcome of the closed books instead of selling the software alone?" 00:58:41

The ratio is $1 software spend to $6 service spend across most enterprise categories. Customer support is already in "autopilot" phase, with Sierra as the leading example.

Reading Founders Is an Active, Vulnerable Practice

Julien's method for founder assessment involves deliberately opening up first — sharing personal backstory including his father's neurological disease, his mother's cancer — to create a psychologically safe environment where founders reveal who they actually are, not just what they've rehearsed.

"My job is in 30 minutes I have to figure out what's special about this person and what might make them exceptional. And I cannot make a mistake because this job is so unforgiving, not when you invest in the wrong company, but when you don't invest in the right companies." 00:31:09

Distance Traveled as a Predictive Signal

Pattern matching on pedigree (school, employer) is a trap. The more powerful signal is how far a founder has traveled from their starting point relative to their peers with identical credentials.

"It just shows you how much distance that person traveled to get where they are. And it doesn't mean that they'll continue on that path. But at least it tells you a lot about their existing trajectory." 00:39:51

The IC Process Is a Signal Amplifier, Not Just a Decision Gate

Sequoia's IC is evolving: async contributions are now layered with the live meeting to capture both fast thinking (verbal debate) and slow thinking (written reflection). Critically, founders still pitch the full IC — and how a founder performs under that pressure is itself data.

"It's a signal. And then you decide as a sponsor what you do with that signal. Did you have questions about the founder being commercial, a good communicator? If you bombed the IC, maybe your questions were founded." 00:27:02

Physical AI and Hardware Require a Different Investment Clock

The shift to hardware and physical AI means the milestone for early validation is no longer ARR growth — it's working prototype. This extends timelines and demands more collaborative, multi-fund capital structures.

"Instead of measuring these companies of how quickly they get from zero to a million in ARR, you have to look at how quickly they get to a working prototype. But you're moving atoms, not bits. So it's just a lot harder." 00:13:14

Cultural Calibration Is a Core Diligence Skill

NPS scores and reference calls must be adjusted for cultural baseline — German customers saying "7" means excellent, while American customers saying "9" may need to be discounted.

"If they're French or German, you add one or two points. If they're Americans, usually you want to retract one or two points on the other hand." 00:38:13

2. Contrarian Perspectives

Neo Labs Are the Quora and StumbleUpon of the AI Era

Most investors are piling into AI model startups (Neo Labs). Julien's private view is that these are likely to be obliterated by the major frontier models, just as social upstarts were wiped out by Facebook.

"I think right now if you're going to invest in a new Neo Lab, you're basically investing in the Quora, in the StumbleUpon when Facebook came about." 00:00:29

The only exception: an N-of-1 founder pursuing a genuinely differentiated architecture (e.g., Ineffable's David Silva, backed by Sequoia at seed).

Don't Invest in Services Companies That Plan to Become Software Companies

Despite Julien's thesis that the services-to-software transition is the next major wealth creation opportunity, he will not back a services business hoping to transform into a software company — because frontier talent won't join, and the data they collect is rarely the right data.

"The best companies are able to concentrate talent. And you're just not going to get frontier talent wanting to work for an old service business that's kind of travestied into an AI company." [00:05:13 — context from 01:04:46]

Founders Who Perform Well in IC Are a Yellow Flag

Unanimous enthusiasm at IC — every partner voting 7 or 8 — is a danger sign, not a green light. The best founders know how to game investor psychology and retrofit their narrative.

"If everyone's a seven or eight, quite dangerous. Because look, founders know what we want to hear. The best founders are able to retrofit the narrative that they think is going to land with investors." 00:30:07

Sequoia's response: force a designated devil's advocate to write the premortem before any investment decision.

Judgment (JQ) Trumps IQ, and Political Coefficient (PQ) Trumps EQ

Sean McGuire's framework adds two dimensions most people ignore: judgment — the ability to find solutions in complex systems — outranks raw intellectual horsepower, and PQ (navigating politically complex systems) outranks emotional intelligence.

"His argument is that judgment is actually more important than IQ. And PQ is more important than EQ." 00:46:11

Billion-Dollar Rounds Are the New Series A

The traditional framing of billion-dollar valuations as "late stage" is obsolete. The same power-law multiple logic that applied to a $50M post-money Series A now applies at the billion-dollar entry point.

"A billion dollars could just be the new Series A. If you think about it, we used to do a $50 million post at Series A hoping it would become a billion. Now you do a billion and it becomes a $20 billion company. Same blunt multiple." 00:11:16

3. Companies Identified

Anthropic Description: Leading AI safety and frontier model company. Why mentioned: Sequoia passed initially, then wrote a $2.5 billion check after updating priors — held up as the canonical example of revisiting a prior and overcoming ego.

"I think we underestimated the company in the early days." 00:10:42

Citadel Securities Description: Ken Griffin's market-making and securities firm; had never taken outside capital. Why mentioned: Sequoia partner Constantine Bueller built a years-long mentorship relationship with Ken Griffin as a student and eventually secured an investment — the paradigm case of relationship-based hunting.

"Constantine never gave up and just kept asking, can we invest? Can we invest? Until Ken kindly said yes." 00:09:24

SpaceX Description: Elon Musk's aerospace and launch company. Why mentioned: Sean McGuire's SpaceX investment received a "1" vote from at least one partner at IC and still became one of the greatest investments in Sequoia's history — the defining case study for individual conviction overriding group skepticism.

"Credits to Sean when he brought in the SpaceX investment. We vote on companies. I think someone voted a one." 00:00:29

Sierra Description: AI-native customer support and customer experience platform. Why mentioned: The clearest current example of the "autopilot" model — charging per ticket resolved rather than per seat, demonstrating software margins on outcome-based pricing.

"Sierra comes in and they say, well, we will resolve those tickets for a fifth of the price... effectively the AI is running the entire workflow end-to-end." 01:00:05

Profound Description: Answer engine optimization (AEO) platform; makes businesses visible to AI chat interfaces. Why mentioned: Described as "the answer to SEO for the modern marketer" — the first mover in agent-facing discoverability, a category Julien sees as enormous.

"They help you make your business visible to people who are using chat interfaces." 00:50:11

Fireworks AI Description: AI inference and model optimization infrastructure company. Why mentioned: Cited as a portfolio company that is "ripping" — first mover in a critical infrastructure layer that benefits regardless of which application-layer company wins.

"They built the best product and they appeal to the best customers. So they're running away with the market." 00:54:18

Relit (Rillet) Description: AI-native finance and accounting platform for software companies. Why mentioned: Highlighted as both a current portfolio example (Sequoia met 17 public company CFOs to drive customer introductions) and an early proof point of the services-capture thesis via "Project Iowa" — their fastest-growing segment is non-tech companies including car washes and auction businesses.

"They have this thing they called Project Iowa. And it's basically appealing to companies outside of tech. And this is the fastest growing segment in the business." 00:55:33

Cursor Description: AI-native code editor; built by Anysphere. Why mentioned: Named the best agent company outside of Sequoia's portfolio — first to understand that you could post-train models and go deeper into the stack, rather than being a pure wrapper.

"They were the first company to really understand that you could post-train models and actually go deeper into the stack." 01:08:34

Harvey Description: AI legal platform for law firms and enterprises. Why mentioned: Cited as the likely winner in legal AI due to widest distribution; Julien is skeptical of the many competitors being funded.

"I think Harvey is very well positioned because they have the widest distribution." 01:07:04

Lagora (Leiga / LaGora) Description: AI-native CRM and go-to-market platform. Why mentioned: Referenced multiple times as an example of hyper-fast ARR growth in greenfield AI markets; Harry backs it and considers it the legal-space analogue winner.

Lovable Description: AI-native app-building platform (formerly GPT Engineer). Why mentioned: Julien's biggest miss — he had lunch with founder Anton Ossika before the company was founded and didn't recognize his potential, teaching him to be more intentional in early founder meetings.

"I had lunch with Anton Ossika from Lovable before he founded the company. I just didn't see it." 00:36:26

Revolut Description: UK-based neobank and financial super-app. Why mentioned: Julien's most painful early-career miss (lost the deal to Index and Balderton); his mother invested at a ~$180-200M valuation and the company is now valued at over $100 billion. Also the deal that taught him the market was not winner-takes-all with Trade Republic.

"I remember thinking this could be bigger than PayPal." 01:11:33

Trade Republic Description: German retail investment and neobroking platform. Why mentioned: Julien's self-identified biggest analytical miss — he told founder Christian it would fail because Revolut would dominate, failing to see it was not a winner-takes-all market.

"I remember telling the founder, Christian, I don't think you're going to succeed because Revolut is going to smoke you." 01:09:07

Ineffable Description: UK-based AI lab pursuing a differentiated model architecture; large seed round. Why mentioned: The one type of Neo Lab Sequoia will back — an N-of-1 researcher with a genuinely different architectural approach, not a "same thing but better" competitor.

"He's an N-of-1 researcher going after a very different type of architecture." 00:12:49

Tacto Description: B2B procurement software for industrial and manufacturing SMEs (Mittelstand). Why mentioned: Julien's first ever Sequoia diligence exercise; used to illustrate the cultural calibration point around German customer reference scores.

Octor (Octo) Description: AI for enterprise software implementation, working with large software vendors. Why mentioned: Both Harry and Julien are investors; cited as evidence that human-in-the-loop implementation work is accelerating (ServiceNow has never hired more system integrators), with Octor enabling one person to do what ten did before.

Granola Description: AI meeting notes and knowledge capture tool. Why mentioned: Harry's biggest personal founder misread at pre-seed — Chris (co-founder) was not articulate in pitch but had extraordinary references; taught Harry to weight references over presentation skills at pre-seed.

UiPath Description: Robotic process automation platform. Why mentioned: Daniel Dines was pitching at the same East London basement event as Nikolai Storonsky (Revolut) — Julien used this to illustrate the extraordinary concentration of exceptional value in a single room.

Odessia (Odyssey) Description: Travel and hospitality AI platform; founded by Francis Davidson (ex-Sonder). Why mentioned: Harry invested alongside Sequoia's Konstantin; used as an example of a sector where agents may soon fully replace human travel planning.

McCaw Description: Data infrastructure company; Harry is an investor. Why mentioned: Cited as an infrastructure bet Harry is more confident in than application-layer companies, alongside Fireworks and ClickHouse.

ClickHouse Description: Open-source column-oriented database management system. Why mentioned: Cited by Harry as a high-conviction infrastructure investment that will benefit regardless of which application layer wins.

Penny Lane Description: European tech company (Sequoia portfolio); Luciana Alessandro investment. Why mentioned: Part of the "banger after banger" pattern attributed to Luciana Alessandro's picking ability.

Stark Description: Accessibility tooling for digital products (Sequoia portfolio). Why mentioned: Cited as a recent Luciana Alessandro investment demonstrating her ability to reinvent across categories.

Framer Description: Web design and publishing platform. Why mentioned: Luciana Alessandro investment at Atomico, cited as part of her consistent cross-category track record.

Deliveroo Description: UK-based on-demand food delivery platform. Why mentioned: Luciana Alessandro's first investment after Julien met her — the opening example of her picking track record.

Solve Intelligence Description: AI for IP law and patent attorneys. Why mentioned: Harry is an investor; used to argue that vertical unbundling within legal will create multi-billion-dollar businesses even within categories that Harvey and Lagora dominate horizontally.

Cloudflare Description: Network security, CDN, and developer infrastructure. Why mentioned: Two contexts: (1) as an example of agent brand bias — agents already default to Cloudflare for hosting; (2) Cloudflare's own prediction of 1,000x agent-to-human traffic in five years.

Vercel Description: Frontend cloud and deployment platform. Why mentioned: Alongside Cloudflare, already a default choice for AI agents selecting hosting infrastructure — evidence of baked-in agent brand bias.

Puzzle Description: Financial reporting and accounting software for startups. Why mentioned: Gloria (from Puzzle) named as the best angel not getting enough credit.

Rocket Internet Description: Berlin-based venture builder; founded by Oliver Samwer. Why mentioned: Julien's first employer out of school; the negotiation with Oliver Samwer over salary led directly to his ability to make angel investments in early Revolut.

4. People Identified

Constantine Bueller (Konstantin) Description: Sequoia partner, based in Europe. Why mentioned: Built a multi-year relationship with Ken Griffin starting as a student; eventually secured Citadel Securities' first-ever outside investment — the firm's example of extreme relationship tenacity.

"Constantine built a relationship with Ken since he was a student. He had been his mentor for years and years." 00:08:55

Sean McGuire Description: Sequoia partner; led SpaceX investment. Why mentioned: Brought SpaceX to IC despite a "1" vote, forced the partnership to visit in person, and created one of Sequoia's greatest investments. Also originator of ELO framework for founder assessment and JQ/PQ dimensions framework.

"He forced all the partnership to fly over to see it with their own eyes. We ended up doing a smaller investment that led to a big investment." 00:17:23

Doug Leone Description: Sequoia senior partner and longtime steward of the firm; Italian-American immigrant. Why mentioned: Embodiment of Sequoia's hunter culture (already on calls before 5am); originator of the "best reference / worst reference" interview question that generates the most honest founder signal.

"He starts by asking, who is your best reference and why? ... And as they finish, you ask the counter question, which is who would be your worst reference and why?" 00:40:52

Alfred Lin Description: Sequoia partner; former COO of Zappos. Why mentioned: Source of the framework distinguishing outlier operators from outlier founders — a critical warning against CV-based pattern matching in the AI era.

"Do not mistake an outlier operator for an outlier founder." 00:45:05

Pat Grady Description: Sequoia partner; leads the firm's growth practice. Why mentioned: "Vectors" framework for founders — direction (why/motivation) × magnitude (ambition/pain tolerance) = predicted trajectory. Also cited for radical humility: "Every company that goes public, we have seen at some point in their journey. That just shows you how many we've missed."

"People are like vectors. And vectors are the product of their direction and magnitude." 00:43:36

Luciana Alessandro Description: Sequoia partner; formerly at Accel; brought Julien into Sequoia. Why mentioned: Named the best "picker" in the partnership — banger after banger across entirely different categories (Deliveroo → Framer → Penny Lane → Stark) with no obvious pattern, demonstrating genuine cross-domain conviction.

"It's just banger after banger. And if you look at the pattern, there's no pattern. It's just across categories." 00:16:12

Dean Meyer Description: Sequoia partner based in Tel Aviv; former professional footballer. Why mentioned: Named the best sourcer in the firm — unique combination of competitive intensity from professional sport and deep technical knowledge; exceptional at connecting with both young spiky founders and serial entrepreneurs.

"He has the competitive juices of Messi, but coupled with the technical depth of someone who's been working in tech their whole career." 00:15:26

Anton Ossika Description: Founder and CEO of Lovable (formerly GPT Engineer); Swedish. Why mentioned: Julien's biggest founder misread — his thoughtful, product-oriented style was mistaken for lack of intensity; the miss taught Julien to be more intentional and prepared when meeting founders.

"I had lunch with Anton Ossika from Lovable before he founded the company. I just didn't see it." 00:36:26

David Silva Description: Founder of Ineffable; N-of-1 AI researcher based in the UK. Why mentioned: Backed by Sequoia at a large seed round as the rare Neo Lab worth funding — pursuing a genuinely differentiated model architecture rather than incrementalism.

"He's an N-of-1 researcher going after a very different type of architecture." 00:12:49

Brian Chesky Description: CEO and co-founder of Airbnb. Why mentioned: Sequoia's Airbnb seed investment — after being turned down by most firms — is cited as one of the highest money-on-money returns in firm history. Noted for his unusual historian's obsession with medieval lodging as a precursor to Airbnb's insight.

Don Valentine Description: Founder of Sequoia Capital. Why mentioned: Source of the "founders you like vs. founders who make money" 2x2 matrix — a foundational framework still used to challenge partners' personal biases in investment decisions.

"If you look at founders you like versus founders who make money as a two by two matrix, your job is to figure out in which part of the quadrant we make money." 00:00:00

Ken Griffin Description: Founder of Citadel and Citadel Securities. Why mentioned: The subject of Constantine Bueller's decade-long relationship-building effort that eventually yielded Citadel Securities' first outside investment from Sequoia.

Gloria (Puzzle) Description: Angel investor; co-founder of Puzzle (financial software). Why mentioned: Named "best angel who doesn't get enough credit" — exceptional nose for companies and works intensely for founders.

"Gloria from Puzzle, I think, deserves a lot of credit. She's got an incredible nose and works extremely hard for her founders." 01:08:34

Nikolai Storonsky Description: Co-founder and CEO of Revolut. Why mentioned: The most obvious founder Julien ever encountered in terms of intensity — Julien camped outside his office to win the deal and still lost it to Index and Balderton; later Julien's mother made a life-changing investment.

"I thought this could be bigger than PayPal." 01:11:33

Daniel Dines Description: Co-founder and CEO of UiPath. Why mentioned: Was pitching at the same East London seed event as Nikolai Storonsky — cited to underscore the extraordinary concentration of founder quality in that single room.

Torsten (Helsing) Description: Co-founder of Helsing, European defense AI company. Why mentioned: Used as an example of elite founders who are almost impossible to get strong positive references from — their standards are so high that even exceptional people merely pass as "okay."

Oliver Samwer Description: Co-founder of Rocket Internet; Julien's first employer. Why mentioned: Julien's difficult salary negotiation with Samwer led to an arrangement allowing angel investing — directly enabling the Revolut investment that made his mother's retirement.

Max (LaGora) Description: Founder of LaGora (AI-native CRM). Why mentioned: Contrasted with Anton Ossika as "an aggressive Swede, an American Swede" — used to illustrate how founder reading must account for national cultural temperament.

Demis Hassabis Description: Co-founder and CEO of Google DeepMind. Why mentioned: Cited in the context of excitement about AI's potential for life sciences breakthroughs — "great to have these big brains focusing on these problems."

5. Operating Insights

Ask "Who Is Your Worst Reference and Why?" — Not Just the Best

Doug Leone's signature interview move generates far more useful signal than the standard reference request. Founders who answer honestly point you directly to where they failed; watching their body language and tempo as they answer tells you even more.

"Who would be your worst reference and why? And see their colors change. People answer that honestly... What's interesting is how they answer the question." 00:40:52

Use Async Writing to Capture Slow Thinking Before Live IC Discussion

Sequoia is now layering written, asynchronous contributions on deal memos before live IC meetings. Live meetings are good for fast, reactive thinking; written contributions capture deliberate, second-order analysis. Getting both is a better decision architecture than either alone.

"An IC is a great format for fast thinking. Speaking asynchronously is great for slow thinking. And so if you can get the benefit of both, you're hopefully going to make better decisions." 00:25:30

Write the Premortem Before You Make the Investment

When an IC runs too smoothly and everyone loves a deal, Sequoia designates a devil's advocate to write the failure case before committing. This counteracts narrative capture by charismatic founders and ensures someone owns the downside scenario.

"We try to write the premortem of that investment before we make it. And we try to spar around that conversation because in a couple of years time, one of us may have to deal with the consequences of that." 00:30:32

Calibrate NPS and Reference Scores by National Culture

Customer NPS and reference enthusiasm must be adjusted for cultural baseline. German customers saying "7" is strong approval; American customers saying "9" may reflect social norm rather than genuine conviction.

"If they're French or German, you add one or two points. If they're Americans, usually you want to retract one or two points on the other hand." 00:38:13

Concentration Over Diversification: Partner With Two to Three Founders Per Year

Julien's operating model is to partner with a maximum of two to three founders per year — sitting in the passenger seat, closing first customers personally, personally meeting CFOs and top hire candidates. This only works with deep concentration, not a diversified portfolio.

"If you look at Relit, we met with 17 public company CFOs since the start of the year. Some of them have become customers. How do you do that when you have 200 companies with 2% in each of them? It just doesn't work." 00:14:48

6. Overlooked Insights

Agent Brand Bias Is Already Being Traded as an Investment Signal

Buried in a paragraph about agents defaulting to Cloudflare and Vercel, Julien dropped a line that reframes a new asset class entirely: hedge funds are already purchasing data on how AI agents make purchasing and tool-selection decisions — because agent preference directly affects the stock prices of the companies being chosen.

"You have hedge funds who are buying data to understand how agents are making decisions because that may influence the stock price of these companies." 00:49:24

This implies that "agent share of mind" is becoming a measurable, tradeable metric — essentially a new form of market research that doesn't yet exist as a structured product but already has institutional buyers. The company that builds the definitive dataset on agent decision-making behavior (not just AEO optimization, but the underlying behavioral analytics) would have extraordinary leverage over both enterprise marketers and institutional investors simultaneously. Profound is mentioned as one answer, but the data layer beneath it is entirely unaddressed.

The "Judgment Is in the Training Loop" Flywheel Is the Durable Moat for AI-Enabled Services Companies

Julien briefly noted that copilot products sitting inside human judgment loops — where an AI watches a human make a complex decision (reading body language, adjusting a question on the fly, sensing hesitation) — are capturing judgment data that cannot be scraped from the public internet and was never in any model's training set. The companies that do this correctly convert today's human judgment into tomorrow's AI intelligence.

"If they're building the right product, they will be able to harness that judgment so that the judgment of today is the intelligence of tomorrow." 01:01:45

This is the actual moat in the services-to-software transition — not the service revenues, not the brand, but the proprietary judgment dataset being quietly assembled inside human workflows. Any company operating as a copilot in a high-stakes, judgment-intensive domain (legal strategy, M&A diligence, clinical diagnosis, complex sales) that is logging the human decision layer alongside the AI assist is building something that cannot be replicated by a competitor starting fresh. This was mentioned in one sentence and then the conversation moved on — but it is arguably the most durable competitive advantage available to any AI company today.