Shruti Gandhi is a solo GP at Array Ventures who uniquely combines being an active AI engineer, Columbia University computer science professor, and Commissioner at the San Francisco Employees' Retirement System, giving her technical, academic, and institutional LP perspectives that are rare among venture capitalists. As the founding engineer of Array, she focuses on enterprise deep tech companies and has achieved 15 realized exits from her pre-seed investments, positioning herself as one of the few solo fund managers with proven early-stage outcomes.

• Enthusiasm for the rise of autonomous AI agents and their ability to execute complex enterprise tasks. Frequently discusses how technology is moving beyond simple screen scraping toward native browsing methods where agents can conduct transactions, such as software that allows bots to go shopping or manage projects without human intervention.
• A sharp focus on the security vulnerabilities introduced by rapid AI adoption. Often warns about the risks of handing production database credentials to language models and highlights specific incidents where deployed agents exposed sensitive API keys or inadvertently launched denial-of-service attacks.
• Strong preference for technical founders who focus on coding and product validation over networking. Explicitly invites cold outreach from builders who spend their weekends writing code, while advising against vague meeting requests or getting trapped in endless fundraising loops without real progress.
• Analysis of how AI reasoning capabilities are dismantling traditional software business models. Points out that general agents are rapidly catching up to vertical-specific tools, suggesting that the standard SaaS approach is vulnerable as models learn to plan, pivot, and execute workflows independently.
• Interest in deep technical infrastructure and hardware efficiency. Commends engineering feats like optimizing inference stacks to run advanced AI on older chips and explores concepts like world models that allow robots to simulate environments before interacting with physical reality.