65labs Unconference - Value-Maxxing is the New Token-Maxxing

​another edition of 65labs unconference - this time for people trying to do more with ai without treating every task like it needs the biggest model, deepest run, or highest spend limit.

​we’ll get into what “getting the most out of your tokens” actually means in practice: how costs stack up across real workflows, where a cheap and fast model is enough, when it’s worth routing to something stronger, and how to think about open vs closed models beyond ideology.

​expect honest examples, not a model leaderboard: routing decisions, spend per task, quality trade-offs, evals, context and retries, and the moments where spending more is actually cheaper because it saves rework or gives you enough confidence to ship.

​we’ll also have folks from the cursor, openai, and google deepmind teams joining us to demo some of the workflows and techniques they’re using in practice.

​speakers:

• ​Gabriel Chua (OpenAI)

• ​Ivan Leo (Google Deepmind)

• ​Agrim Singh (Cursor/SpaceXAI)

• ​Community demos - ​Rach Pradhan, Phil (from Airfoil), and Swathi Bhat.

​come if you’re building, researching, or just getting started in ai!

​location: carousell

​doors open at 5:50pm

​note: if you would like to submit a community demo - no slides pls - feel free to reach out to sherry@65labs.org. we will not be taking any company pitches or sales presentations - only educational content.