Letronna: Local AI agent with cloud tasks

Letronna is an AI agent that can work on a computer or run tasks on the publisher's cloud. It offers desktop downloads and a terminal interface. The main decision for a user is whether a job should run on the local machine or on Letronna's servers.

What is available on a computer?

The publisher lists desktop builds for macOS, Windows and Linux. Letronna describes working with files, commands and web search, along with editable memory between sessions. Its documentation covers how to see and change that memory. For jobs involving project files, review what the agent intends to read or change before approving an action.

Letronna says the local agent can use a user's own model keys. An account provides access to hosted models. The distinction matters when choosing which provider will receive a request or how model usage is billed.

What changes when a job runs in Letronna Cloud?

The publisher says a cloud task can keep going after the desktop app or laptop closes. It can give a web app a preview address, or keep a service running as a worker. Letronna describes different behavior for these resources: a web app can sleep after inactivity, while a worker can continue running and incur compute charges. The cloud documentation also says conversations' folders are removed after 14 days of inactivity unless a worker remains active.

What is included in the free and paid options?

Letronna lists local use as free. Hosted models are billed from credit, and Letronna Cloud is part of Pro, with cloud compute drawn from credit as well. Using an existing model account does not mean the provider's own usage is free. If a task needs to stay online, check the cloud billing rules and sleep behavior before using the preview as a permanent service.