Delta Podcast Ep. 53: Brian Zhan, Partner at Striker VP and Investor in Reflection, Skild, Periodic, Ricursive
Dec 23, 2025
Brian Zhan, partner at Striker Venture Partners and former product lead on Facebook’s Presto team, brings deep expertise in data systems, robotics, and AI-for-science investments. He discusses open-sourcing MPP systems, early bets on vector search, backing robotic foundation models, and investing in AI-driven bioscience and research startups. Practical startup advice and frontier research trends round out the conversation.
AI Snips
Chapters
Transcript
Episode notes
Product Truth Beats Benchmark Gaming
- Early category conviction and shipping a demonstrably better product often trump benchmark wars.
- Brian argues customers seeing a superior free trial adoption beats engineered benchmark wins like those by incumbents.
Dyna Focused On Practical Robotics Readiness
- Brian backed Dyna Robotics based on founders Joel and Abhinav who pioneered robotics foundation models and applied transformer-style architectures to control.
- The team focused on building end-to-end product readiness: hardware fixes, cheaper sourcing, and large-scale data collection for imitation learning and RL.
Reflection Chose Open Weights For Enterprise Agents
- Brian met Misha and Yannis through referrals and invested in Reflection for AI digital actions and open-weight enterprise models.
- Reflection built on the team's expertise (behind Gemini) to offer open weights enabling stronger in-house engineering agents for enterprises.