Nandu Anilal writes detailed public research on AI and data infrastructure through Canaan posts and his notes & upside Substack, giving founders unusual transparency into his technical thinking on areas like data engineering and ML observability. He brings direct product strategy experience from Thumbtack's marketplace team and uses a user-research approach to evaluate companies, asking founders about end-user needs rather than just market sizing.

• Bullish outlook on AI-native software development and its potential to unseat incumbents. He specifically mentioned testing tools like Bolt and v0 which convinced him that digital product creation is shifting, while questioning how established design giants like Figma will maintain their narrative against the rise of code generation tools.
• Analysis of winning strategies in crowded or seemingly commoditized markets. He often looks at companies that succeeded despite late entry, such as Monday.com growing into a massive business years after Asana, or ServiceNow proving that software defensibility is still very real with high retention rates despite narratives to the contrary.
• Focus on emerging cybersecurity threats and enterprise security dynamics. He highlighted a specific concept called slopsquatting where AI hallucinations create fake code packages that attackers exploit, and he closely follows market moves from major players like Wiz and the competitive history between Axon and Flock Safety.
• Interest in the intersection of sports and technology, particularly within the NBA. Beyond just watching the playoffs and the Pacers, he critiqued the current inefficiency of the referee review system and predicted that the league will inevitably adopt AI-driven reviews to minimize game disruptions.
• Observations on polarizing startup operational models in a post-pandemic world. He noted a distinct split in successful hiring philosophies, seeing founders choose between either small, intense in-person teams in expensive hubs or larger, remote-first global teams, with very little middle ground remaining effective.