Ontologies, Evals & Deployed Intelligence: Building Real AI Agents (Basis CEO)

How do you actually build AI agents that work in the real world? Mitchell Troyanovsky, Co-Founder of Basis, goes deep on what it takes to deploy agents in accounting — one of the hardest and most high-stakes domains.

In this talk, Mitch breaks down:

Why ontology (a coherent world for agents) is the missing piece of agent design

How evals at scale separate production agents from prototypes

What it means to build a deployed intelligence team inside a company

The tradeoffs between human memory vs. agent context management

How to think about applied research in agents (context compression, environment engineering, memory systems)

Subscribe for more deep dives with AI founders building at the frontier.

This session was recorded live at a recent Data Driven NYC, our in-person, monthly event series. If you are ever in New York, you can join the upcoming events by following FirstMark on Luma: https://luma.com/firstmarkcap

Basis
Website - https://www.getbasis.ai
X/Twitter - https://x.com/trybasis

Mitchell Troyanovsky
LinkedIn -   / mitchelltroyanovsky  
X/Twitter - https://x.com/mitch_troy

HOSTED BY:

FIRSTMARK
Website - https://firstmark.com
X/Twitter -   / firstmarkcap  

Matt Turck (Managing Director)
LinkedIn -   / turck  
X/Twitter -   / mattturck  

Check out the MAD Podcast:
Spotify - https://open.spotify.com/show/7yLATDS...
Apple - https://podcasts.apple.com/us/podcast...