Venture capital firms are relying on AI, but flawed data sources are undermining investment decisions. Henrik Landgren argues that direct access to company financials and better data infrastructure could transform how investors spot and evaluate startups.
Venture capital investing has always balanced intuition with analysis, but as AI tools become more common in the industry, the quality of underlying data is emerging as a critical issue. Henrik Landgren, co-founder and CPTO at Gilion, contends that many firms are missing the mark by focusing on AI adoption without first addressing the data that feeds these systems.
Landgren, who previously served as Spotify’s first VP of analytics, recalls how granular user data revolutionized decision-making at the streaming giant. In contrast, he found the investment world lagging behind, still relying on founder-supplied data that is often curated to present the best possible narrative. This information asymmetry, he says, is a fundamental flaw in the venture capital model.
Today, AI is often used to quickly summarize pitch decks or generate reports, but Landgren warns that these efficiencies are superficial if the data itself is incomplete or biased. “Garbage in, garbage out” remains true: large language models and other AI tools can only be as effective as the information they process.
Landgren advocates for a shift in approach. Instead of relying on data packaged by founders, investors should connect directly to primary sources-such as payment systems, accounting software, and board reports-to gain a more accurate picture of a company’s operations. This would allow analysts to focus their expertise on evaluating teams and business models, rather than spending time cleaning and verifying data.
Improved data access could also help investors identify promising startups that are often overlooked, such as capital-efficient businesses or those in less trendy sectors. With better infrastructure, venture capitalists could move faster and with greater confidence, potentially beating competitors to the most attractive deals. As the landscape evolves and new categories like AI-powered hardware and deep tech emerge, traditional metrics may no longer suffice, making robust data pipelines even more essential.
Landgren’s perspective echoes concerns raised in recent coverage of the sector, where the rush to back AI startups has sometimes led to inflated valuations and increased risk, as seen in reports like this analysis of multi-tiered funding rounds.
Ultimately, Landgren argues, the question is not how AI can speed up existing processes, but how the entire investment workflow can be rebuilt around reliable, direct data. Without this foundation, he warns, venture capital will continue making high-stakes bets with limited visibility-just at a faster pace.
