🎙️How I Invest: The AI-Native VC…

1. Key Themes

The Asymmetric Cost of False Negatives in Venture

In a power-law-driven asset class, missing a winner is far more costly than backing a loser — and this asymmetry should reshape how firms structure their investment process.

"If you invest €1M into a startup and it fails, you lose €1M. If you don't invest €1M into a company that would have returned 50x, you effectively miss €50M of potential upside... false negatives are incredibly expensive."

The Lovable case makes this concrete: Earlybird passed at pre-seed — partly because available ownership (~8%) was below their 15% target — and at a $13B valuation the investment would have "likely returned north of 50x and potentially become a fund-returning investment by itself."

Proprietary Institutional Data as the True Source of AI-Driven Alpha

As foundation models and data providers commoditize, the durable advantage shifts to a firm's own accumulated "institutional exhaust."

"Differentiation moves somewhere else: proprietary data. And by proprietary data, I don't necessarily mean some secret external dataset. I mean your own institutional exhaust. Meeting transcripts. Investment memos. IC decisions. Partner ratings. Rejection reasons. Founder interactions. Portfolio performance."

Earlybird operationalized this with a structured post-IC survey introduced in 2018 that tracks partner judgments against actual outcomes — enabling questions like "Which partners are unusually good at evaluating teams?" and "Where are we systematically too pessimistic?"

Context-Dependent Founder Assessment Across the Technology Stack

There is no universal "great founder" archetype — the ideal profile is a function of where a company sits in the technology stack.

"The deeper you sit in the technology stack, the more important capital becomes. The higher you sit in the stack, the more important distribution becomes."

At the Deep Tech layer, technical credibility and manufacturing knowledge dominate. At the frontier AI layer, talent gravity is paramount ("there might only be a few dozen exceptional researchers globally"). At the application layer, "product velocity, customer understanding and distribution matter more" because "building software has become dramatically cheaper."

AI Should Concentrate Investor Attention, Not Expand Volume

The correct use of AI productivity gains is depth, not breadth — moving investors further down the funnel rather than evaluating more companies.

"If AI makes investors 10x more productive, should we evaluate 10x more companies? No... The objective shouldn't be to replace it with more work. It should be to move the investor further down the funnel."

Earlybird receives 10,000+ inbound opportunities per year but makes only 10–12 investments. AI handles top-of-funnel filtering so investors can shift from "four relatively shallow founder meetings" to "five deeply prepared conversations."

Change Management — Not Technology — Is the Binding Constraint on AI Adoption

The hardest part of building an AI-native firm is behavioral transformation, not technical implementation.

"One of my biggest misconceptions was assuming data would be the hardest problem... The harder problem turned out to be change management."

When Earlybird built EagleEye (a sourcing platform improving European deal coverage from ~71% to 95%+), the majority of the team initially didn't use it despite its obvious superiority. The solution was to measure outcomes (coverage rates by geography) rather than mandate tool usage: "Don't force people to use new tools. Measure the outcome you want and give them better tools to achieve it."

2. Contrarian Perspectives

The VC industry's biggest contradiction: it was built to back unconventional ideas, yet increasingly clusters around consensus.

"Venture capital was created to finance unconventional businesses that traditional capital providers wouldn't touch. Yet investors increasingly cluster around the same companies... FOMO becomes rational."

The evidence: Earlybird's 2015 seed investment in UiPath out of Romania — "when raising money for the company was still difficult" — returned at a ~$33B IPO valuation six years later. The opportunity was exceptional because it wasn't consensus. Chasing consensus both inflates prices and misses the best opportunities simultaneously.

More AI productivity should lead to fewer, higher-conviction decisions — not more activity.

The instinct when productivity increases is to do more. The article argues the opposite: once an investor has sufficient pattern recognition, "more information can become noise" and the "scarce resource becomes cognitive capacity."

"Early in your career, quantity matters enormously. You need reps. Thousands of companies. Thousands of decisions... But once you have accumulated those reps, more information can become noise. Then the scarce resource becomes cognitive capacity."

This runs against the prevailing narrative of AI enabling hyperactive, high-volume deal sourcing.

The dashboard is dying as the primary interface for data-driven investing.

"For years, data-driven VC meant building dashboards. I increasingly believe the dashboard is dying. The future interface is conversational. Instead of: Human → Software → Database, we increasingly move toward: Human → AI → Everything."

This implies that firms still investing in dashboard-centric analytics infrastructure are building toward an already-obsolete paradigm.

3. Companies Identified

Earlybird VC

Description: European early-stage VC firm with $2.5B AUM
Why mentioned: Primary case study for rebuilding an investment firm around AI and data over nearly a decade; built proprietary sourcing platform EagleEye and internal AI workflows
Quote: "What we learned over almost a decade of transforming Earlybird into an increasingly AI-native investment platform"

UiPath

Description: Enterprise automation software company; Romanian-founded, IPO'd in 2021
Why mentioned: Exemplar of a non-consensus investment that generated outlier returns; Earlybird invested at seed in 2015
Quote: "Earlybird invested in UiPath out of Romania in 2015. Our Romanian Partner Dan Lupu led the seed round when raising money for the company was still difficult. Six years later, UiPath went public at roughly $33B."

Lovable

Description: AI-powered application builder; valued at ~$13B at time of writing
Why mentioned: Cited as a costly anti-portfolio miss — Earlybird passed at pre-seed partly due to insufficient ownership available (~8% vs. 15% target)
Quote: "At their recent $13bn valuation, the investment would've likely returned north of 50x and potentially become a fund-returning investment by itself."

EagleEye (Earlybird internal tool)

Description: Proprietary sourcing platform built by Earlybird
Why mentioned: Concrete example of AI-native deal sourcing infrastructure; improved European deal coverage from ~71% to 95%+, typically six weeks before financing rounds close
Quote: "After building EagleEye, we increased this to roughly 95%+, meaning we now see approximately 19 out of every 20 relevant opportunities, typically at least six weeks before their financing round closes."

Granola

Description: AI meeting copilot that transcribes from device audio (no meeting bot)
Why mentioned: Newsletter sponsor; notable for privacy-first, bot-free architecture that works across all meeting tools and in-person
Quote: "Granola transcribes directly from your computer or phone audio. It works across any meeting tool: Zoom, Google Meet, Microsoft Teams."

4. People Identified

Andre Retterath

Description: Partner at Earlybird VC; author of the Data Driven VC newsletter
Why mentioned: Author and primary voice of the article; architect of Earlybird's AI-native transformation since joining in 2017
Quote: "When I joined Earlybird in 2017, I encountered something that surprised me. Some of the smartest people I'd ever met were spending enormous amounts of time doing repetitive manual work."

David Weisburd

Description: Host of the How I Invest podcast
Why mentioned: Interviewed Andre Retterath; posed the central question about where VC alpha comes from when all firms access the same AI models
Quote: "This leads to perhaps the most important question David asked me: If every VC has access to the same AI models, where does alpha come from?"

Dan Lupu

Description: Romanian Partner at Earlybird VC
Why mentioned: Led the seed investment in UiPath in 2015, exemplifying high-conviction, non-consensus investing in non-obvious geographies
Quote: "Our Romanian Partner Dan Lupu led the seed round when raising money for the company was still difficult."

5. Operating Insights

Measure outcomes, not tool adoption, to drive behavioral change.

When Earlybird wanted its team to use EagleEye, mandating usage failed. The solution was to measure deal coverage by geography as a performance metric. Since EagleEye provided 95%+ coverage, it became the path of least resistance to hit the measured outcome.

"Instead of telling people to use EagleEye, we started measuring the outcome: coverage... But when EagleEye contained 95% of relevant opportunities, the easiest path toward the desired outcome became obvious."

Applicable beyond VC: any organization adopting new tools should define the performance metric the tool serves, not the tool itself.

Build AI on top of first-principles process redesign, not workflow augmentation.

Earlybird didn't ask "where can we use AI?" — they asked "if we built an investment firm from scratch today, how should it work?" The distinction produced a fundamentally different system architecture.

"We didn't start by asking: 'Where can we use AI?' We asked: 'If we built an investment firm from scratch today, how should it work?' That's a very different starting point."

For operators: AI implementations that merely accelerate existing processes will capture a fraction of the value available to those who redesign the process around AI capabilities.

Structure the week to protect cognitive white space as a deliberate resource.

"Monday through Thursday are generally meeting-heavy. Parts of Friday remain blocked for reading papers, newsletters and research. Friday provides input. The weekend provides space to process. Monday returns to execution. Execute → Learn → Think → Execute. AI should create more room for this cycle, not fill every newly available minute with another task."

6. Overlooked Insights

Early-round valuation discipline can systematically eliminate the best investments.

The Lovable miss wasn't caused by poor judgment about the company — it was caused by a portfolio construction rule (minimum ~15% ownership) that mechanically filtered out the opportunity. This points to a structural risk in how many funds operate: rigid ownership thresholds designed to optimize portfolio math may systematically exclude the highest-upside, most competitive deals.

"We saw the company at pre-seed and passed partly because the roughly 8% ownership available was below our fund target of around 15%... The lesson isn't to ignore valuation or ownership. It's to continuously ask whether your investment rules are helping you find outliers — or systematically filtering them out."

Founder brand is evolving from a "nice to have" into durable competitive infrastructure, especially as building costs collapse.

This point receives less emphasis than deal sourcing or AI workflow, but carries strategic weight: if software development costs continue falling and 50 teams can build technically similar products, differentiation migrates almost entirely to distribution, speed, and brand. Founder branding then becomes a force multiplier that compounds across hiring, fundraising, and customer acquisition simultaneously.

"If 50 teams can build technically similar AI products, differentiation increasingly moves toward distribution, talent, speed and brand. In that environment, founder brand becomes part of the company's infrastructure."