From Vibe Coding to Agentic Engineering | DeepMind, Mastra, Constructive (Happy Hour + Panel)

Software creation just mass-democratized. Now what? Forty-one percent of all code written globally is now AI-generated. Over 90% of US developers use AI coding tools daily. In a single year, the market for AI-native development tools crossed $4.7 billion. This is one of the most important shifts in the history of software. And it raises a question that doesn't get enough serious attention: When anyone can create software, what determines whether that software is actually worth running in production? The answer isn't more code generation. It's engineering — the kind that lives beneath the application layer, in the data infrastructure, the authorization model, the trust architecture, and the systems-level thinking that separates a demo from a product. This panel brings together leaders from Google DeepMind, Mastra, and Constructive — spanning frontier AI testing, open-source agent frameworks, and database infrastructure — to explore what engineering actually means in the age of AI-generated software. The Questions We're Here to Answer How is the practice of software engineering actually changing? Not the marketing version. The real version. What does the day-to-day look like for engineers building with and on top of AI systems? What new skills matter? What old skills matter more than ever? Where are the sharpest teams placing their bets? What does the full production stack look like for agentic software? From frontier models to data layers to application frameworks — what's load-bearing and what's hype? When an AI agent needs to read sensitive data, make a decision, and write back to a database, what has to be true about every layer of the stack for that to be safe? Where does trust live in AI-native architecture? Agents are probabilistic. Business logic is deterministic. Authorization can't be approximate. As AI systems gain more autonomy — making database calls, mutating state, accessing multi-tenant data — where do you enforce the rules? In the application? The model? The infrastructure? What did the vibe coding wave actually unlock — and what's being built on top of it? The ability to generate software from natural language isn't the end state. It's the foundation. The most interesting companies in this space are already building the next layer: agentic development environments, AI-native workflows, and tools that make AI-generated software reliable at scale. What does that next chapter look like? Where is value migrating as code generation becomes commoditized? When the act of writing code gets dramatically cheaper, the premium shifts to everything around it — architecture, security, data integrity, trust, and the ability to ship safely at speed. Who captures that value, and how? Panelists Sanjeev Dhanda Sr. Staff Software Engineer, Google DeepMindSanjeev leads the infrastructure shift toward non-deterministic and agentic testing for Gemini at Google DeepMind. He co-authored the chapter on Continuous Integration in O'Reilly's Software Engineering at Google and his ICSE paper on continuous testing at Google scale has been cited over 450 times. Dan Lynch Founder & CEO, Constructive Dan is building secure-by-default Postgres for the agentic era. His open-source database tooling runs in production across 10 million+ databases at companies including Supabase and Databricks, with over 100 million downloads. When agents query data autonomously, the permission model changes fundamentally — Dan is building the trust layer that makes that safe. Abhi Aiyer Founder & CTO, Mastra Abhi is Founder and CTO of Mastra, the leading open-source TypeScript agent framework — YC-backed with $35M raised including a $22M Series A led by Spark Capital. Mastra powers production agent systems at Replit, PayPal, and Brex. Previously Abhi built Gatsby, one of the defining frontend frameworks of the last decade. Why This Panel Is Different This is a technical and strategic conversation between people who are building at different altitudes of the same stack — and who see the same transition from different vantage points: The frontier AI lab rethinking how you test systems that don't behave the same way twice. The agent framework that developers actually use to ship autonomous systems into production. And the database infrastructure company building the trust layer that holds it all together. The conversation between these three perspectives is where the real insight lives. Each panelist represents a layer of the stack. Together, they map the full architecture of what it means to build software in 2026 — and what has to change as we move from generating code to engineering systems. Who This Is For This event is for people who are past the novelty of AI-generated code and are now dealing with the real questions: How do I ship this safely? How do I architect for agents? Where does authorization live? What breaks at scale? How do I build trust into the system, not just the interface? If you're an engineering leader, founder, CTO, infrastructure architect, platform engineer, or investor tracking the AI-native software stack — this is the room you want to be in. Details Date: Friday, May 1st, 2026 Time: 5:30 – 8:30 PM Format: Happy hour + panels + networking SVB Experience Center 532 Market Street, San Francisco (corner of Market & Sansome) The iconic rotunda space at 1 Bush Plaza — one of SF's premier venues for the tech and innovation community. The Evening 5:30 – 6:30 PM — Happy Hour + Networking Kick off the weekend right. Drinks, bites, and a room full of engineers, founders, and operators who are building at the frontier of AI-native software. This is the part where you meet the person whose problem sounds exactly like yours. 6:30 – 6:35 PM — Welcome A brief scene-set from your hosts. 6:35 – 7:15 PM — Panel: From Vibe Coding to Agentic Engineering A focused conversation with engineering leaders from Google DeepMind, Mastra, and Constructive — three perspectives across the full AI-native stack on what it actually takes to go from AI-generated prototype to production-grade agentic software. 7:15 – 7:30 PM — Audience Q&A Open floor. Come with questions — the people on stage are building the infrastructure that the next generation of software runs on. This is your chance to pressure-test their thinking. 7:30 – 8:30 PM — Networking + After-Hours The conversations that start at the bar after the panels are usually the best ones. Stick around. Hosted By Frontier Syndicate Frontier Syndicate is a private venture circle for high-caliber founders, investors, and frontier technologists. The syndicate combines curated convenings with special-purpose investment vehicles to help exceptional companies attract aligned capital, strategic relationships, and long-term advantage. Silicon Valley Bank Silicon Valley Bank (SVB), a division of First Citizens Bank, is the preferred bank of the world's most innovative companies and investors. SVB provides commercial banking to companies in the technology, life science and healthcare, private equity and venture capital industries. SVB operates in centers of innovation throughout the United States, serving the unique needs of its dynamic clients with deep sector expertise, insights and connections. SVB's parent company, First Citizens BancShares, Inc. (NASDAQ: FCNCA), is a top 20 U.S. financial institution with more than $200 billion in assets. First Citizens Bank, Member FDIC. Sponsored By Constructive is building secure-by-default Postgres for the agentic era — a platform that moves trust, authorization, and multi-tenant control into the database layer so teams can ship AI-native and enterprise-grade applications faster and more safely. Built by a team behind 100M+ open-source downloads, Constructive is making the database the trust boundary for the next generation of software. Acasia builds and manages private AI cloud infrastructure — sourcing dedicated GPUs (H100, B200, MI300X and beyond), securing power and data center capacity, and delivering a managed platform so teams can train, deploy, and scale without the cost and complexity of assembling it themselves. Acasia is the single-source infrastructure partner for companies that need GPU capacity to scale.