# Map the work.

> Design multi-step agent operations as understandable flows of skills, tools, approvals, policy gates, data transformations, and completion evidence.

Canonical: https://rangoon.ai/product/workflows/

Status: In active development. Product screenshots are design previews with illustrative data. No general availability is implied.

## A graph with semantics

Planned workflow nodes can represent triggers, agents, skills, tools, branches, waits, subflows, exception handling, and audit events.

### Meaningful connections

A data flow, required dependency, approval gate, and tool invocation should read as different relationships.

## Govern the consequential moments

Policy and approval gates belong within the operational path where people can see what is requested and why.

### Receipts and states

Completion evidence and outcome states are designed to be explicit rather than implied by a chat transcript.

## Start simple, extend carefully

The workflow experience is being designed for understandable small flows as well as more advanced orchestration.

### Preview status

Workflow editing and runtime behavior remain in active development.


## More information

- [Documentation index](https://rangoon.ai/llms.txt)
- [AI access and policies](https://rangoon.ai/ai/)
- [Source repository](https://github.com/hypler-dev/rangoon)