Why Your Next Production System Doesn't Need a GPT
Frontier LLMs are powerful — but they're expensive, slow, and require an internet connection. This talk makes the case for Small Language Models (SLMs): models under 10B parameters that run entirely on your laptop, deliver real production results, and cost $0.00 per inference. We cover the full stack — from local inference runtimes to multi-agent pipelines — with five live demos running on Apple Silicon.