Skip to main content
Rilo is a managed AI service that completes knowledge work end-to-end. You delegate to it in plain language, and it decides how to handle each message. Knowing how Rilo handles a message — and how it does the work — helps you delegate effectively.

Quick answers vs. queued work

When you message Rilo, it first decides what kind of response your message needs:
  • Quick answers. Greetings, simple questions, and things Rilo can answer from context come back immediately, in the same conversation.
  • Queued work. Anything that needs Rilo to use your tools, gather information, or take several steps becomes a task that Rilo plans and executes. You’ll get progress updates and a result when it’s done — see Following Up.
You don’t have to label which is which; Rilo routes your message for you. Writing a clear request helps Rilo plan the right work — see Writing Requests.

How Rilo does the work

For queued work, Rilo plans a sequence of steps, then carries them out using its skills across your connected tools. It observes what happens at each step and adjusts — if something doesn’t work, it re-plans rather than blindly retrying. Outputs are grounded in the task’s actual context and sources, not guessed.

Rilo learns as it goes

Rilo remembers your organization’s preferences, context, and what worked and applies that to later tasks. The same task tends to get faster and more aligned the more Rilo does it. This memory is shared across your organization and across every channel you use to reach Rilo.

You stay in control

  • Rilo works with least privilege — standard user-level access through secure connections, not broad admin rights. See Security & Privacy.
  • Some requests are out of bounds on purpose — Rilo won’t make account, billing, or security changes, and tells you where to go instead. See What Rilo Can’t Do.
  • Everything is partitioned by organization — your data, memory, and connected tools are never shared across organizations.

Where to go next