AIoT foundations
What is AIoT? Where artificial intelligence meets the Internet of Things
AIoT combines connected devices with machine learning. Learn what it means, where the intelligence runs, and when it is worth the extra complexity.
AIoT Report
AI meets Things
Where machine learning leaves the data center and runs next to the sensor. Edge AI, TinyML, on-device vision, predictive maintenance and the agents that are starting to operate homes and factories.
This section is also published as AIoT Report and has its own RSS feed.
Photo: Jainath Ponnala / Unsplash
Edge AI
Edge AI explained: running machine learning on IoT devices
How edge AI works: why inference moves out of the cloud, the hardware tiers, how models are shrunk with quantization and pruning, and when to avoid it.
TinyML
TinyML: machine learning on microcontrollers
TinyML runs neural networks on microcontrollers with kilobytes of RAM. Learn the constraints, the workflow, the frameworks and the workloads that fit.
On-device vision
On-device computer vision for IoT
How to run computer vision on IoT hardware: the capture-to-action pipeline, task types, hardware tiers, model choices and the limits of local processing.
Predictive maintenance
Predictive maintenance with IoT and AI
How predictive maintenance works: vibration and current signatures, the signal processing behind them, the models that use them, and where it pays off.
AI agents
AI agents for the smart home
How LLM-based agents control a smart home through Matter, Home Assistant and MQTT, what guardrails they need, and when a plain automation is better.
AI agents
AI agents on the factory floor
How LLM agents fit into a plant: reading OPC UA and MQTT data, drafting work orders and diagnoses, and the guardrails that keep them away from control loops.
AI terms in the glossary
All AI terms- AI agentSoftware that pursues a goal by observing its environment, reasoning about what to do and calling tools or devices to act, often built on a large language model.
- AIoTThe combination of artificial intelligence with Internet of Things systems, so that connected devices and their data pipelines can learn, classify and decide rather than only measure and report.
- Anomaly detectionIdentifying readings or patterns that deviate from what a system has learned to consider normal, without needing examples of every possible fault.
- Computer visionThe field of extracting information from images and video with software, from reading a gauge to tracking people and vehicles.
- Condition monitoringContinuously measuring parameters such as vibration, temperature or current on a machine to track its health and detect deterioration early.
- Deep learningA branch of machine learning that uses neural networks with many layers to learn features directly from raw signals such as images, audio or vibration.
- Edge AIRunning machine learning inference directly on or near the device that produces the data, instead of sending everything to a cloud service first.
- Federated learningA training approach where devices improve a shared model locally and send only parameter updates to a server, so raw data never leaves the device.
- Generative AIModels that produce new content, such as text, images, code or synthetic sensor data, rather than only classifying or predicting from existing inputs.
- InferenceThe act of running a trained machine learning model on new data to produce a prediction, classification or score.
- Keyword spottingA small always-on audio model that listens for a specific word or phrase, such as a wake word, and triggers a larger process only when it hears it.
- Large language model (LLM)A neural network trained on very large text corpora that can understand and generate language, and increasingly also call tools and interpret structured data.