Local-first · Open source desktop
Persistent Sage keeps chats and Memory Anchors in local SQLite. You bring your own API keys, choose your provider, and decide what leaves the device. No Persistent Sage cloud stores your conversations.
Memory Anchors — durable facts and insights, not another forgetful chat tab.
Your provider — OpenAI, Anthropic, Gemini, xAI, OpenRouter, or local Ollama.
Desktop free forever — MIT open source. Microsoft Store and GitHub installs.
Three modes, one local desktop app. Companion for continuity, Productivity for day-to-day surfaces, Coding for repos—plus Moltbook and Google when you opt in.
Companion
Multi-thread conversations with Memory Anchors in the sidebar, project chips above the composer, and chat artifacts—charts, HTML, tables, and forms—rendered right in the thread. Favorites, Share, and copy actions sit on every reply.
Productivity
Movable widgets for Gmail, Calendar, Tasks, Contacts, Drive, Projects, weather, and more. Connect Google once; tokens stay encrypted on your machine. Layout resets and add-widget controls keep the board yours.
Coding
Manage git repos under workspace/repos with a file tree, multi-tab editor, allowlisted terminal, and sandboxed playgrounds for Python, Node, Markdown, and JSON. The coding agent can search, patch, and run commands when you enable those tools.
Moltbook
Browse Hot, New, Top, and Rising while your registered agent posts, comments, and upvotes—on a schedule you control. Humans configure voice and boundaries; the agent does the interacting.
Google Workspace
Sign in with Google for Productivity widgets and companion tools. Connect a separate agent mailbox when you want correspondence handled on its own thread. Gmail, Calendar, Drive, Contacts, and Tasks stay scoped to what you enable.
Personality
Shape tone, values, and special instructions—or import OpenClaw markdown and JSON profiles. A live system prompt preview shows what the model actually sees before you save.
Persistent Sage is a privacy-oriented desktop companion. Multi-thread chat, long-term memory, customizable personalities, optional agent tools, and a coding workspace—without handing your history to a Persistent Sage server.
Compact long-term memory separate from raw chat. Hybrid keyword, full-text, and optional semantic recall injects what matters into new sessions.
Conversations live in SQLite on your machine. API keys are encrypted at rest. There is no Persistent Sage-operated cloud for chat storage.
Use OpenAI, Anthropic, Google Gemini, xAI Grok, OpenRouter, Ollama local or cloud, or an offline placeholder. Switch when your needs change.
Multiple companion profiles with tone, values, and instructions. Import OpenClaw markdown or Persistent Sage JSON and keep continuity across threads.
Flip between chat continuity, a Google-backed desk of widgets, and a repo-scoped IDE with terminal and playgrounds—linked to the same companion when you want it.
Web search, browser fetch, workspace files, PDF tools, Moltbook, and Google—all off by default until you enable them.
Precision is the marketing. When you use a cloud LLM, that provider receives the content you send. Local Ollama keeps inference on your machine. We say both out loud.
Chat history, Memory Anchors, personalities, settings, and attachments remain on disk unless you deliberately send content to a provider or tool.
Prompts and images go to the AI provider you select. Optional web tools contact third-party sites. Update checks hit GitHub or the Microsoft Store.
Persistent Sage does not operate a backend that stores your chats, memories, or API keys. No ads. No telemetry SDK to a Persistent Sage server.
Grab the Windows installer from GitHub Releases, or look for Persistent Sage on the Microsoft Store. Desktop is free forever. Mobile Free + Premium is coming.
