local ai assistant
Local AI Assistant for Mac
Local-first, not fully offline. GuideLayer runs on your Claude Code and Codex sign-ins before any metered API fallback.
"Local AI" gets used for everything from fully offline Llama installs to chat apps that happen to have a Mac client. The details matter.
GuideLayer is local-first: it runs on your Mac, stores encrypted data on your Mac, and routes inference through your Claude Code and Codex sign-ins before any metered cloud fallback.
Not air-gapped. Not sending your screen to a GuideLayer server. Your machine, your sign-ins, your Privacy Center settings.
What actually works
Claude Code and Codex first
Primary inference path uses subscriptions already on the device. No extra per-token bill from GuideLayer for daily use.
Data stays on device
Saved tutorials, encrypted conversations, and Privacy Center prefs live locally. Not a vendor cloud account.
Apple Silicon native
Signed, notarized Mac app built for Apple Silicon. Menu-bar native, not a web wrapper.
Optional API fallback
Metered cloud APIs available when you need them. Not the default path for every query.
What local-first means here
Fully offline AI (Ollama, LM Studio) keeps everything on device but can't match frontier models for complex screen reasoning. Cloud-only assistants get the models but want your context in their cloud. GuideLayer sits between: frontier models through your own sign-ins, screen context captured locally on demand, no GuideLayer server owning the pipeline.
Workflow examples
Daily assistant without a new subscription
You already pay for Claude Code or Codex. You don't want another AI meter.
- 01Sign in with what's on the Mac.
- 02Use voice and screen guidance through that path.
- 03API fallback only when you explicitly need it.
Tutorials stored on the machine
Team walkthroughs shouldn't live in a vendor's cloud by default.
- 01Record and save tutorials locally.
- 02Export Markdown or PDF to share however you want.
- 03Replay from the Mac without re-uploading to a platform.
Evaluating vs Ollama or cloud chat
You're deciding between local models, cloud chat, and something screen-aware.
- 01Ollama: best for offline text/code, no screen guidance.
- 02Cloud chat: best for reasoning, weak on seeing your UI.
- 03GuideLayer: screen-aware, local-first storage, your model sign-ins.
Local AI on Mac isn't one thing. For screen guidance with your own model sign-ins and no vendor in the middle, that's the lane GuideLayer occupies.
Related guides
Try GuideLayer on your Mac
Screen-aware, private, local-first. Download for Apple Silicon with Claude Code and Codex already signed in.