this is my role model

I cannot forgive Descartes. In his whole philosophy he would like to dispense with God, but he could not help allowing Him a flick of the fingers to set the world in motion, after which he had no more use for God -Blaise Pascal Pascal Knows

great event joining the AI community in Miami! definitely drop by if you're around I'll be speaking remotely on the winning workflow for llm evals and presenting a nice internal tool we built to find interesting traces from your production data https://t.co/RaXsMLmM4C

Built an experimental fork of Deep Agents @LangChainAI that lets you spin up deep agents, connect them to MCP servers (Claude-style), with pre-built tools and specialized subagents from a simple CLI. An MCP registry that allows you to: - Connect to remote MCP servers using the

my biggest ai engineering finding this month is that Cursor's auto model selector is actually quite good and this might save me $100 ide's for the masses are all about multi model and routing if you can 10x decrease the cost for big teams you win

its hard to explain what being Argentinean means but this gets very close

the N1

@FieldDataInc @LangChainAI

Code:

Eager to hear the community’s feedback and how this can be enhanced to fit other data review workflows!

I'm not at all convinced this is the optimal way of doing this, and I know as a fact there are many useful features that this is missing. This is just a proof of concept we build for ourselves, to aid in our own data review process.

This is why I came up with this UX. It reunites the best of both worlds: models helping you 'soft cluster' the data in real time, while you define which are the clusters you care about. Moving seamlessly between llms sorting results and traditional search.

Do I want to chat with my traces? Sort of, but I also want to manipulate them and pick a few for review / sharing with the team. Do I want a model to help me surface up interesting patterns? Yes, but I also want to search for more examples myself, using full text search.

What is the best way to explore new patterns when analyzing thousands of traces? Definitely not to go over each one of them. Something along the lines of going broad, then deep, then broad again. And picking up the golden nuggets I find in the process to add to my datasets.

I've always felt torn on how to do discovery on user behavior. I like traditional filters but sometimes I just want to explore patterns in a more unstructured way, randomly strolling through the data until I find something worth diving into.

Sharing an overview of the trace reviewer we developed at @FieldDataInc , thank you to the @LangChainAI team for the feedback! Some key points: * soft clustering of traces (promptable) * reasoning model running in bg * hybrid UX for discovery Code and more details in 🧵:

Our hearts are restless until they rest in you, O Lord

any kind of agentic moat is obliterated by better models

Que pedazo de evento se viene este viernes 🔥 Keynotes de @ideasrapidas, @IvannaAFigueroa, @gptcrosa, @shroominic, @HeyFardo, @manuelsoria_, @fpingham, @lgesuelli_p, @PetralliLucas & more en las oficinas de @lemonapp_ar 🫡 📌Full schedule: https://t.co/uJn9jQwOUB

the fact that people will watch Friends when The Office exists is enough to disprove the efficient market hypothesis