Skip to main content
Ara runtime visual
Your own persistent AI runtime: files, terminal, tools, and messaging access.

​
What every user gets

  • Persistent personal runtime: your assistant keeps working context over time.
  • Isolated sandbox: separate filesystem and process space per user.
  • Real tools: file browsing/editing, terminal execution, web fetch/search, and integrations.
  • Phone-first access: message your assistant through messages (RCS/SMS and iMessage where available).
  • Always-on behavior: workflows and automations can continue beyond one chat turn.

​
How a request flows

1

You ask

You send a message from the app or phone channel.
2

Ara runtime plans

The agent decides whether to answer directly or use tools.
3

Tools execute

Work happens inside your sandbox: files, commands, web, and connectors.
4

You get results

The assistant replies with output and keeps context for future tasks.

​
Under the hood

LayerRole
app.ara.soUser interface and control surfaces
Cloud APIAuth, session lifecycle, routing, integrations
Sandbox runtimeIsolated execution environment per user
Agent runtimeTool loop, memory, and automation behavior

​
What this unlocks

  • Long-running task execution instead of one-shot answers
  • Project-style workflows with files and terminal access
  • Personal context continuity over time
  • Practical automation with the Python SDK

SDK Quickstart

Build your first automation fast.

Reference

Full SDK command and API surface.