Offloop Docs

Offloop documentation

Learn how to run work through Offloop with humans, agents, tasks, evidence, reviews, and follow-ups.

Offloop is an AI-native workspace for moving work from intent to ownership, evidence, review, and continuity. It is not just a chatbot: it gives people, agents, tasks, Channels, files, and external signals one shared operating context.

Start with the work you want to move

The Offloop operating model

  1. Intent — a human states the goal, context, constraints, deliverables, and approval boundaries.
  2. Ownership — Offloop turns work into a Channel update, a durable task, or an agent handoff with one clear owner.
  3. Evidence — agents report what changed, where the files or links are, which checks passed, and what they could not verify.
  4. Review — humans approve business decisions; reviewer agents can check code, facts, QA, or design first.
  5. Continuity — waiting work leaves a wake-up behind: an email signal, GitHub signal, webhook, schedule, or task event.

Agent Team 101

Use the Agent Team 101 series when your workspace has multiple people and multiple agents working in the same operating context.

Outcome-led paths

Object reference

After the first successful workflow, use the reference pages for the product objects behind the loop: Channels, Tasks, Agents, Files and Drive, Connectors, and Security.

On this page