Making the surreal, real. Join us for Surreal in San Francisco, a dinner series held at art galleries.

Everyone is building in AI. Very few know what actually matters.

In the past few years, the playbook for AI startups has completely changed. New primitives (LLMs, agents, and retrieval) have unlocked new possibilities, but also created new traps.

So what actually works today?

In this conversation, we bring together two founders with very different vantage points:

Will Bryk, founder of Exa, is building AI-native infrastructure for search and retrieval which is designed for how models, not humans, access information.
Deon Nicholas, founder of Espa and formerly co-founder/CEO of Forethought (acq. Zendesk), built and scaled an AI company before the LLM wave and is building a new AI native company.

Together, we’ll go deep on:

What “AI-native” actually means in practice (beyond just calling an API)
Where startups still have real technical leverage—and where they don’t
The biggest mistakes founders are making right now
What’s actually defensible in an AI startup in 2026
Where the next opportunities are for builders

This is a technical, candid conversation for people serious about building in AI.

Past dinner attendees have hailed from Humans&, Inflection AI, Perplexity, OpenAI, X.Ai, Reflection AI, Sierra, Glean,, Vercel, Anthropic, Mercor, Midjourney, Harvey, Google Deepmind, Pika, Jane Street, Nvidia, Parallel Web Systems, 1X, Sesame, Stanford CS PhDs, and more.

Agenda:

5:15pm Mingling
6:00pm Conversation
6:40pm Dinner
7:30pm Mingling

Surreal

Surreal is a monthly dinner series & technical community run by Premise.

The series is held at art galleries as we believe that technologists are inspired by creative environments when they invent new technology products, and that the same creativity that inspires artists, inspires technologists.

Learn more about the Surreal community here: surrealbuilders.com

Thank you to our sponsors: Fenwick, SVB, Colliers, & True Search.

Surreal Dinner Series: What Actually Matters When Building an AI Startup