Izzy
1.6K posts
Izzy
@isidoremiller
merry wanderer of the night, AI research @ Hex
For now on call me SEX!!!
what i find most interesting about the decomposition angle is that it treats reasoning capacity as a property of the decomposition formalism around the model. like the limiting factor is often whether the system lets the model express only shallow, explicitly enumerated subcalls or whether it can instantiate richer computational structure (recursion, loops, reusable subroutines) that can efficiently represent exponentially larger task graphs as depth grows. imo this is a very important shift in perspective bcoz it suggests that capability jumps may come less from scaling the base model and more from expanding the space of admissible decompositions while keeping each local call in-distribution.
🎙️Introducing Max Agency Max Agency is a new podcast where we go deep on how the best agents are actually being built: architecture decisions, tradeoffs, evals, and everything in between. Each episode, I sit down with engineering leaders who are doing this work in production. Our first episode features Izzy Miller (@isidoremiller), AI Engineer at Hex (@_hex_tech). Hex has been shipping data agents since before most teams were even thinking about them, starting with single-cell text-to-SQL and graduating to a full Notebook agent that can work autonomously for 20 minutes on a complex analysis. Izzy has a lot of perspective on what it actually takes to get agents working well in production, and what breaks along the way. A few takeaways from our conversation: - Keep your eval sets small enough to hold in your head: Izzy runs 30-50 handcrafted "traps" with multiple repetitions, rather than hundreds of variants. If you can't explain why your agent fails each one, your eval set is too big - Day zero performance is almost irrelevant: The more interesting question is how the agent compounds. Izzy is building a 90-day simulation where the warehouse evolves and the agent has to accumulate understanding - You can catch agent errors without seeing the raw outputs: By running an LLM-as-a-judge over production usage and clustering the results, you can surface places where something likely went wrong, without needing to read individual conversations Watch the full episode on: - Youtube: youtube.com/watch?v=Xyh1Eq… - Apple Podcasts: podcasts.apple.com/us/podcast/how… - Spotify: open.spotify.com/episode/1BJlg3…
Made seventy six clams！！！ I've never seen this Death Valley map blanket in person. Years ago I got a req for a map blanket of this area, so I tried this 1934 promo map and told the weaver to send it directly to the custie. They loved it, sent pics, four more sold since :^)
🎙️Introducing Max Agency Max Agency is a new podcast where we go deep on how the best agents are actually being built: architecture decisions, tradeoffs, evals, and everything in between. Each episode, I sit down with engineering leaders who are doing this work in production. Our first episode features Izzy Miller (@isidoremiller), AI Engineer at Hex (@_hex_tech). Hex has been shipping data agents since before most teams were even thinking about them, starting with single-cell text-to-SQL and graduating to a full Notebook agent that can work autonomously for 20 minutes on a complex analysis. Izzy has a lot of perspective on what it actually takes to get agents working well in production, and what breaks along the way. A few takeaways from our conversation: - Keep your eval sets small enough to hold in your head: Izzy runs 30-50 handcrafted "traps" with multiple repetitions, rather than hundreds of variants. If you can't explain why your agent fails each one, your eval set is too big - Day zero performance is almost irrelevant: The more interesting question is how the agent compounds. Izzy is building a 90-day simulation where the warehouse evolves and the agent has to accumulate understanding - You can catch agent errors without seeing the raw outputs: By running an LLM-as-a-judge over production usage and clustering the results, you can surface places where something likely went wrong, without needing to read individual conversations Watch the full episode on: - Youtube: youtube.com/watch?v=Xyh1Eq… - Apple Podcasts: podcasts.apple.com/us/podcast/how… - Spotify: open.spotify.com/episode/1BJlg3…
