Losant MCP for AI. Control your entire industrial IoT ecosystem from chat.

Losant gives your AI agent full command over connected hardware and smart infrastructure. You talk to your agent, and it executes commands against your whole IoT platform.

Managing Your Infrastructure
You can start by listing all active applications with list_applications, or you can pull the full metadata package for a specific app using get_application. If you need to make an exact copy of an existing setup to test something out, run clone_application; conversely, if an application is dead weight, you can delete it entirely with delete_application.

You'll also manage data structures by initializing a new table with create_data_table, or viewing all tables tied to an app using list_data_tables. Need to clean up? You can remove selected records from a data table rows using delete_data_table_rows.

For device management, you'll add a new physical or simulated unit with create_device, and check the current operational status for any single ID via get_device. To see your whole fleet, run list_devices; when you’re done with a piece of hardware, you delete its record using delete_device. You can also update non-operational metadata—like changing a device's location or owner—using update_device, and if an app needs tweaks but shouldn't be deleted, use update_application.

Building and Controlling Workflows
You define complex operational sequences using create_workflow to build out multi-step logic, like a maintenance routine. The agent can then retrieve the full step-by-step definition of that workflow with get_workflow, or view all defined workflows for an app by calling list_workflows. If you need to adjust the sequence later on, use update_workflow.

To execute a workflow's logic without relying on standard triggers—say, forcing the next step in a safety protocol—you press a virtual button via press_virtual_button. You can also remove an entire defined workflow using delete_workflow.

Interacting with Data and Sensors
When you need to know what's going on, your agent handles it. To find out the single most recent reading for any sensor or metric, call query_last_value. If you gotta deep dive into history, pull a full time-stamped log over a date range using query_time_series, or run a targeted query against a data table to pull only records matching specific criteria using query_data_table_rows.

You can also simply view all visible rows from a data table immediately with get_data_table_rows. To save your findings, you export all relevant device readings into a downloadable file using export_data, or generate and download a full backup of an application's configuration via export_application.

For quick analysis, the platform lets you run a pre-built data analysis notebook by executing it with execute_notebook. You can also get a list of all available notebooks ready to go using list_notebooks, and if you need a summary view of your system status, you fetch the configuration and current state of a dashboard using get_dashboard, or see every configured dashboard name with list_dashboards.

Sending Commands and Reporting
This is where the action is. You send an immediate command—like telling a valve to open or rebooting a unit—to a specific device using send_device_command. If you need to test something out or log a state manually, you push that value directly with send_device_state, and if you want to simulate a physical button press for testing purposes, use the dedicated tool.

You can also change what your agent reports about a device's connection status—Online or Offline—via set_device_connection_status. To generate an official report based on a dashboard’s current state, run send_dashboard_report.

Core Utilities and File Handling
For general visibility, you can list all uploaded files attached to an application using list_files, or upload new documents or data files into the app's storage area with upload_file. You also manage your underlying resource base by getting a list of every available file (list_files), and retrieving a simple summary of all applications connected via list_applications.

If you need to perform foundational system tasks, you can get a full rundown of what's in an app with get_application, or use the general update commands: for example, listing notebooks via list_notebooks and then running them if needed. You don’t have to write custom API scripts; everything is built into these callable tools.