← DocumentationDeployment

Cloud or local AI agent: which deployment should you choose?

Compare virtual cloud agents with local or on-device deployments across privacy, cost, reliability and maintenance.

A cloud agent is usually the fastest way to begin. A local agent can provide greater control for suitable workloads, but it introduces hardware and maintenance responsibilities. The right choice depends on the data, integrations and availability the workflow requires.

When cloud delivery fits

  • The agent needs to be available from anywhere.
  • It relies on cloud email, calendar, CRM or messaging services.
  • You want simple upgrades and managed hosting.
  • The workload changes frequently or needs access to several models.

When local delivery fits

  • Suitable information should remain on a device or private network.
  • The agent works heavily with local files or internal systems.
  • Some functionality must continue without an internet connection.
  • The organisation can manage the required hardware and access controls.

Local does not automatically mean offline

A local agent may still call cloud models, messaging services or external integrations. A genuinely offline design requires local models, local storage and local tools, which affects hardware requirements and the capabilities available.

A hybrid approach

Many practical deployments keep sensitive documents and retrieval locally while using approved cloud services for selected reasoning or communication. Each data path should be documented so the customer understands what leaves the device.

The decision checklist

  • Classify the information the agent will process.
  • List every service it must connect to.
  • Decide the required availability and recovery process.
  • Compare hardware, support and model-usage costs.
  • Document where data is stored and processed.
Contact