Good Companies Research LLM
Research happens in the desktop app.
The local model needs native setup, so the Research LLM is a Forge native desktop app. The macOS download is a drag-to-Applications DMG; once installed, the app can check GitHub releases, show an Update pill, install Ollama, prepare the recommended model, open mission webpages, extract source text, stream synthesis, and keep drafts review-gated.
What the app shows
- Live webpage lens: a main research window loads each candidate webpage as the run visits it.
- Phase moments: short mode shifts such as Scoping, Reading, Vetting, Extracting, Thinking, and Ready.
- Thought bubbles: brief research observations appear as the run extracts text or finds uncertainty.
- Output sidebar: mission brief, webpages, scratchpad notes, and draft JSON stay visible and scrollable.
- Local setup: Ollama install, service start, model install, and readiness checks live in Settings.
Why it is native
The website can list public missions and accept reviewed submissions, but it should not own local runtime setup. The desktop app can install tools, start local services, store local state, and show the research process without asking the browser to manage Ollama directly.