-
RAG Is Dead. Most Banking Apps Have Shitty Chatbots
Traditional RAG can't handle real-world complexity — especially multilingual users navigating foreign banking systems. Agentic RAG is the fix.
-
Vercel's agent-browser: Why I Prefer It Over Playwright MCP
A Rust-powered browser automation CLI built for AI agents — no Node.js, no Playwright, just CDP and a clever 'skills' system that teaches your AI assistant how to use it.
-
Anthropic's Tool Search Tool: Free Your Context Window
MCP is dead? But we need it. Anthropic's Tool Search Tool fixes the context bloat problem — not with compression, but just-in-time selective loading of tool definitions.
-
7-Step NeMo Guardrails' OpenCLAW Enterprise Strategy
NVIDIA's NeMo Guardrails provides a 7-step enterprise framework to sandbox the entire lifecycle of an AI agent — from preflight checks to policy enforcement.
-
Google releases Gemini 2.0 Flash with native tool use
Gemini 2.0 Flash is Google's fastest multimodal model yet, now with native code execution, function calling, and grounding — all in a single API call.
-
AG2: Why I switched from AutoGen to its community fork
AG2 (the community continuation of AutoGen) is what I'm using to build the PulseOS agentic framework at CEGID — here's why.
-
LLM evals are the new unit tests — and most teams are skipping them
Running an agentic evals team at CEGID has made one thing clear: shipping LLM-powered features without evals is like shipping code without tests. Here's our approach.