Generate IoT solutions from plain English, detect anomalies with AI models, and connect AI agents directly to your ThingsBoard data.
This section covers the AI tools available in ThingsBoard — from generating new solutions to opening the platform up to external AI agents. AI for IoT in ThingsBoard works at every stage of a project. Each tool is documented separately below.
Designed for full control over your solution from the command line — creating, updating, and extending it over time. Working alongside Claude Code, describe any change in plain English and let the agent apply it. Commit your configuration to git like any other codebase, and roll the same solution out to dev, QA, and production with a single command.
Learn more about ThingsBoard CLI →Designed for creating new solutions from scratch directly inside the ThingsBoard UI. Describe your use case in plain English — “Track temperature in cold storage facilities and alert on spoilage” — and it generates all entities and dashboards, then deploys everything straight to your tenant. Works with new configurations only — to edit existing dashboards, alarm rules, or calculated fields, use the AI Assistant.
Learn more about AI Solution Creator →Designed for creating and editing calculated fields, and alarm rules inside ThingsBoard UI. Describe what you need in plain English — generate a new alarm rule or adjust a calculated field — and the assistant will help. Available directly on each relevant page inside the ThingsBoard UI.
Learn more about AI Assistant →Plug OpenAI, Google Gemini, Anthropic, Azure OpenAI, Amazon Bedrock, or any OpenAI-compatible provider into your tenant. Predict future values, detect anomalies, classify device states, or generate natural-language insights from live telemetry — invoked from any rule chain via the AI request rule node.
Learn more about AI Models →Run inference on-premise with self-hosted Ollama models — keep sensitive telemetry inside your perimeter and avoid per-token costs.
Learn more about Local AI with Ollama →Build an end-to-end IoT anomaly detection pipeline — devices stream vibration, temperature, and acoustic data, calculated fields maintain a rolling window, and an AI model classifies anomalies in real time. Alarms and notifications fire automatically.
Learn more about Predictive Maintenance →Expose your ThingsBoard tenant as a Model Context Protocol server. Talk to your devices, telemetry, alarms, and entity relations from Claude Desktop, Claude Code, Cursor, VS Code Copilot, or any MCP-compatible client — no REST calls or dashboard clicks required.
Learn more about MCP Server →Orchestrate ThingsBoard from n8n workflows: trigger alarms when an AI model detects an anomaly, push telemetry to a data warehouse, sync assets to a CRM, or chain ThingsBoard data into a multi-step AI pipeline — without writing custom integrations.
Learn more about n8n node →What AI providers does ThingsBoard support?
ThingsBoard integrates with OpenAI, Google Gemini, Anthropic, Azure OpenAI, Amazon Bedrock, and any OpenAI-compatible API. For on-premise inference, ThingsBoard also supports self-hosted models via Ollama — Llama 3, Mistral, Gemma, and others run directly on your own servers.
Can I use AI in ThingsBoard without writing code?
Yes — the AI Solution Creator and AI Assistant let you generate solutions, dashboards, alarm rules, and calculated fields from plain English inside the ThingsBoard UI, with no code required.
How does ThingsBoard connect to AI agents like Claude?
Via the MCP Server: expose your ThingsBoard tenant as a Model Context Protocol server and query or control it from Claude Desktop, Claude Code, Cursor, or VS Code Copilot using natural language — no REST calls required.
Which AI features are available in Community Edition?
AI Models (including Ollama), the AI request rule node, the MCP Server, the n8n node, and the Predictive Maintenance pipeline are available in all editions. The AI Solution Creator, AI Assistant, and ThingsBoard CLI require Professional Edition or Cloud.
