The 50x that got away: Earlybird's Andre Retterath on discipline, data and the Lovable miss
Key takeaways from How I Invest interview with Earlybird Venture Capital General Partner Andre RetterathDealroom has a profile for this one. Try Dealroom → (August 2025):
The Lovable miss. Retterath passed on Lovable at the pre-seed round because Earlybird's stake would have been only ~8%, below the fund's 15% average ownership threshold. He estimates the miss cost north of 50x — likely a fund-returner.
Asymmetric cost of discipline. His framing: lose money and you lose it once; skip a winner and you forgo multiples of the upside. A rejected deal never appears on the track record, making the invisible false-negative the most dangerous — and most repeatable — mistake in venture.
Fundraising is a skill that compounds. Earlybird's own research shows founders who raise at higher valuations with less dilution keep doing so round after round. The deals filtered out on price are disproportionately the ones worth chasing, so the firm has selectively loosened its ownership discipline.
An AI-native firm. Retterath joined in 2017 as an investment professional (not an engineer) and built internal tooling that scaled to eight full-time engineers at peak. The result is EagleEye, a proprietary sourcing platform now covering ~95-96% of relevant European opportunities at least six weeks before rounds close — up from ~72% in 2019.
Measure what matters. Junior investors are now assessed on geographic hit rate rather than tool usage, which finally drove EagleEye adoption. The firm attributes deal sources to cohorts, found large investment conferences underperformed, and pivoted toward vertical AI research events like ICML and ICLR.
The stack dictates the trait. Deep/hard tech founders need capital-attraction skill; frontier AI labs need the gravity to recruit scarce research talent; application-layer founders win on distribution and unit economics. Earlybird splits its team accordingly across deep tech, AI infrastructure and applications.
Founder branding as leverage. Retterath argues a personal brand is a scalable way to package a founder's ability to sell a vision — increasingly unavoidable higher up the stack where vibe-coded lookalikes make differentiation hard.
Culture, not data, was the hard part. He expected data quality to be the barrier; instead it was change management — getting experienced investors to unlearn old workflows. Today the whole firm runs on Claude Code with internal skills and MCP-connected data, managing ~€2.5B in assets.
Read more: Data-Driven VC
