AI that works on prayers ⛪️

If you've tried building AI agents via AgentExecutor in LangChain, you know how it can end.

You give the LLM 15 tools, write a giant system prompt, and hope it doesn't mess anything up. And on simple tasks it works. But as soon as we do something complex — the model calls tools in the wrong order, forgets context, or just goes into an endless loop of hallucinations. 🫠

We don't need an artificial intelligence that "decides for itself what to do." We need a rigid deterministic pipeline where the LLM is just one of the data processing nodes.
This is exactly the problem LangGraph solves.

🗓 May 15 at 6:00 PM Moscow time we will break down this base at Tochka Sborki.

The broadcast and recording will be available to participants. Access is arranged through the bot: https://t.me/TScompiler_bot