AI-native personal knowledge base for macOS

Your files, AI-native.

Connect notes, PDFs, docs, webpages, images, and code, then ask or delegate with context.

One place for files, context, and agents.

Guace turns your local files into an AI-native knowledge base. Ask across formats, inspect sources, and send follow-up work to the right agent.

Use notes, PDFs, documents, webpages, images, code, and plain text without converting everything to markdown.

Chat across your personal knowledge base and inspect the source file beside the answer.

Send coding, writing, or research tasks to managed local workers when an answer needs real execution.

No hosted workspace to migrate into. Guace sits on top of local folders, provider keys, and AI workers you choose.

Point Guace at your personal library. It reads common file formats, keeps source context, and avoids a markdown-only system.

Dispatch a task, keep the transcript, inspect the workspace path, and decide what should be saved back to your files.

Use the providers and keys you choose. Guace routes requests straight to your configured backend.

See real token counts, model cost, and prompt-cache behavior instead of guesswork.

Start with OpenAI, Anthropic, Vercel, and DeepSeek. Add or swap providers when your work changes.

Your knowledge base starts as files on disk. Your keys live in Keychain. When a model is needed, Guace talks to the provider you configured.

Portable sources. Your documents stay searchable, inspectable, and useful without locking them into a hosted workspace.
Provider choice. Requests go to your selected backend with your own credentials.
Worker isolation. Managed agents run in task workspaces instead of writing directly over your knowledge base.
Cost visibility. Token counts, model spend, and cache behavior are visible inside the product.