Cloud architecture, with a clearer story.

A practical workflow

1. Define the audience and the system boundary. Separate users, edge services, application services, and storage.
2. Name your provider and the components you actually use. Add queues, caches, identity boundaries, and external services where relevant.
3. Label important connections and inspect the result. Check ingress, data access, trust boundaries, and the failure paths before sharing.

Try this starting prompt

Draw an AWS web application: users reach CloudFront, then an application load balancer and two application instances. The application uses RDS PostgreSQL and Redis. Show a private data tier and label request and database connections.

Adapt the names, connections, and constraints to your own system. AI output is a draft: inspect its assumptions and edit the result before sharing.

Choose the handoff

Use the visual canvas for presentation, Code or Split view for precise changes, and saved versions before major revisions. A read-only link is useful for a quick review; export source or an editable project backup when the recipient needs to continue editing.