TSTanisha Soniintanisha-soni.hashnode.dev·Jul 5 · 9 min readI Taught My Local Model to Speak LangChainTwo articles in, and both of them assumed something: that you're using a model LangChain already knows how to talk to. OpenAI, Anthropic, whatever, doesn't matter. LangChain has an adapter for it. Wha00
TSTanisha Soniintanisha-soni.hashnode.dev·Jul 4 · 4 min readYour LLM Doesn't Need to See the Token Either Last article, I showed you how to stop shoving 47 tool descriptions into your LLM's context window. This one's about a different problem: what happens after the LLM picks a tool. Here's the thing nobo00
TSTanisha Soniintanisha-soni.hashnode.dev·Jul 3 · 7 min readYour LLM Doesn't Need to Know About All 47 ToolsI built a support agent that loaded 40+ tools from MCP. The agent kept getting confused, latency crept up, and my token bill wasn't pretty. The fix turned out to be a FAISS index show the LLM only the00
