AI 4UAnalyze my business

Plain-language AI glossary

Term 34InfrastructureMeaning / context / connections

Infrastructure / Definition

Edge AI / On-Device AI

Running AI models directly on user devices (phones, laptops, IoT) rather than sending data to cloud servers for processing.

Browse all termsUnderstand the idea
34of 75
01

MeaningThe one-sentence definition.

02

ContextHow the idea works in practice.

03

UsesWhere the concept becomes useful.

01 / Plain-language context

How Edge AI / On-Device AI works.

Edge AI processes data locally on the device, eliminating the need for an internet connection and keeping data private. Apple's Core ML, Google's MediaPipe, and frameworks like ONNX Runtime enable running optimized models on mobile devices and laptops. Apple Intelligence on iPhone runs smaller models entirely on-device.

The benefits are compelling: zero latency (no network round-trip), complete privacy (data never leaves the device), offline capability, and no API costs. The constraints are equally real: limited model size (phones have far less memory than GPU servers), lower accuracy compared to cloud models, and battery/thermal considerations.

For mobile app builders, edge AI works well for: image classification, object detection, text autocorrect, voice commands, and simple text generation. Complex tasks like multi-turn conversation, code generation, or long document analysis still require cloud models. Many apps use a hybrid approach: edge AI for quick, private tasks and cloud APIs for complex ones.

02 / Practical uses

Where it helps.

  1. 01Offline AI features in mobile apps
  2. 02Privacy-preserving AI processing
  3. 03Real-time camera and sensor analysis
  4. 04Low-latency voice and gesture recognition
03 / Keep exploring
InferenceThe process of running a trained AI model to generate predictions or outputs from new inputs, as opposed to training the model.QuantizationA technique that reduces AI model size and memory requirements by using lower-precision numbers to represent model weights, trading a small accuracy loss for major efficiency gains.DistillationA technique where a smaller "student" model is trained to replicate the behavior of a larger "teacher" model, achieving comparable quality at lower cost.LlamaMeta's open-weight model family, available for download and deployment across cloud, data-center, edge, and local environments under its license.