Build313
← Own Your AI

Advanced AI Build Night · Local Models

Before you start.

Have these ready before the session so you can build along from minute one.

  • 01

    A machine with enough RAM

    16 GB is the sensible minimum (runs 7–8B models comfortably). 8 GB works for small models only; 24 GB+ opens up 20B+ models. Apple Silicon or a recent GPU helps but isn't required.

  • 02

    Ollama installed

    Free, macOS / Windows / Linux. Install, then it serves an OpenAI-compatible API at localhost:11434.

    Open ↗
  • 03

    OpenWork installed

    Free, open-source desktop app (MIT). macOS and Linux are first-class; Windows support is newer. No terminal needed to use it.

    Open ↗
  • 04

    One model pulled ahead of time

    Downloads are multi-GB. Run `ollama pull qwen3:8b` (or llama3.1:8b) before the session so you're not waiting on Wi-Fi.

  • 05

    A folder to point an agent at

    A real working folder — notes, a few PDFs, a spreadsheet. This is what your local agent will read and act on.

  • 06

    API keys for hybrid mode (optional)

    Only if you want to fall back to a frontier model (OpenAI / Anthropic) for the hard parts. Local-only needs no keys at all.

  • 07

    An Ollama account (we follow along on the cloud)

    Run `ollama signin` before the session so you can run glm-5.2:cloud live — the free tier is enough.

    Open ↗