Purpose and Positioning Spatio is a workspace-centric AI agent platform (Agent Harness), paired with a workspace control plane. It unifies scattered work artifacts such as notes, spreadsheets, slides, emails, tasks, and chats into one place, driven by agents and kept in sync. Users can bring their own agents (Claude Code, Cursor, Codex, or any MCP client), or directly use the agent built into the Spatio desktop app.
Main Capabilities
- Workspace agent: The official description states that Spatio Agent is an agent execution framework based on wafer-level inference, running at 1851 tokens per second; when a task requires more complex processing, it can hand the task off to the Claude Code or Codex the user has already subscribed to, and merge the results back into the same thread.
- Unified workspace view: The home page centralizes the entire workspace in a sortable, dense table, supporting simultaneous filtering, grouping, and browsing across platforms without switching between multiple apps.
- Nestable structure: Groups serve as nestable containers for any work artifact, with optional templates, allowing users to readjust the structure according to how their team actually works.
- Context awareness: What the user currently has open is passed along with the conversation, eliminating the need to repeatedly explain the project, re-paste documents, or repeatedly @-mention the same files.
- Data access methods: Native connectors are provided for deeply integrated systems, third-party connectors for long-tail systems, and Spatio Browser for platforms without reliable APIs, letting agents operate in real accounts and real sessions.
How to Use
- Desktop: Spatio Desktop provides a unified workspace interface, with the built-in agent available for direct use.
- CLI: Supports operating the workspace via the command line. The official examples given are as follows: spatio -p "…" — one-off headless run, prompt input, streaming reply output to stdout, convenient for piping, scripts, and CI. spatio chat — addressable session runs, opening a session, sending turns, tracking the event stream as NDJSON, and passing the session ID between processes. spatio do <action> — execute platform operations from the shell, such as completing tasks, sending emails, creating notes; new platforms are automatically connected without updating the CLI.
- MCP interface: Spatio Workspace Agent Harness is exposed through MCP, allowing Claude Code, Codex, Cursor, or other clients supporting the protocol to drive it, thereby gaining access to all tools and real data within the workspace.
- Cross-device continuity: Work started on the desktop can continue running in the cloud, and full context is restored upon return.
Applicable Scenarios Suitable for teams that need to uniformly manage work artifacts across multiple platforms and accelerate daily tasks with AI agents. For example: summarizing Gmail threads, drafting replies, updating Sheets rows, posting summaries to Teams, etc. (the official example is about 2600 output tokens). It is also suitable for developers who integrate agent capabilities in the terminal via CLI or MCP clients.
Limitations and Information Not Provided Official materials do not mention specific pricing, fees, platform compatibility scope, deployment requirements, or privacy policy details. The inference speed, token rate, and similar figures mentioned in the text are officially published data, and actual performance may vary depending on the environment.