Local OS and Cloud OS
Consuelo OS can run locally or in the cloud. The distinction is about runtime, storage, sharing, and observability. The product model is the same: agents connect to the OS portal, run approved skills, create tracked artifacts, and leave an observable trail.Comparison
| Capability | Local OS | Cloud OS |
|---|---|---|
| Runtime | User machine | Hosted in the regular OpenSaaS deployment path |
| Storage | Local filesystem and SQLite | App database and cloud storage |
| Best for | Solo users, founders, internal builders, local agent workflows | Teams, shared workspaces, customer deployments |
| Agents | Local connected agents | Local or hosted agent connections, depending on setup |
| Artifacts | Local bytes plus SQLite metadata | Cloud bytes plus app-native metadata |
| Approvals | Local first, team surfaces later | App, chat, or team workflow surfaces |
| Observability | Local logs and trace records | Hosted logs, app records, and observability tools |
| Setup complexity | Low | Higher, but team-ready |
Local OS
Local OS should not feel like an untracked script runner. It should create the same kinds of records cloud OS creates, just stored locally. Local OS should include Bun, SQLite, an OS home folder, connected agents, local artifact storage, execution logs, trace IDs, and capability health. Local is the fastest way to build and use OS before every hosted capability is wired.Cloud OS
Cloud OS is the team version. It should include workspace and user identity through the app, app-native artifact records, cloud artifact storage, shared visibility, hosted observability, managed capabilities, and approval delivery surfaces. Cloud OS is the version that makes OS sellable for teams.Choose local when
- you are the only user
- you want fast setup
- your agent runs on your machine
- you are building or testing a new skill
- you need local files or local scripts
Choose cloud when
- multiple people need access
- artifacts need to be shared
- workspace data is already in the app
- capabilities should be managed server-side
- approvals should go to team surfaces
- support and observability matter