Categories

Artificial Intelligence

Phi-3 on the edge: Microsoft’s SLM in 2025

Phi-3 is Microsoft Research's family of small language models aimed at the edge, and it competes directly with Llama 3.2, Gemma 2 and Qwen 2.5. Phi-3-mini holds 3.8B parameters and, quantized to 4 bits, fits in about 2 GB, running on a CPU with a neural accelerator, an integrated GPU or an NPU.

Industry 4.0

Industrial Edge Computing: Low Latency Where Data Happens

Industrial edge computing moves processing capacity from the centralised cloud to the plant floor, the machine, or the robotic cell. Local latency (10-50 ms) is critical for process control, machine vision, and safety systems: it is a physical limit that bandwidth alone cannot solve. OPC UA, K3s, and private 5G now form a proven production-ready stack.

Technology

Cloudflare Workers: Edge Compute Without Containers

Cloudflare Workers run JavaScript and WebAssembly across more than 300 cities with no servers to manage and cold starts close to zero. Where the model genuinely shines, which workloads it handles badly, and the constraints worth knowing before you commit to it.

Software Development

WASI 0.2: The New Standard Interface for WebAssembly

WASI Preview 2 redefines how WebAssembly interacts with the operating system. It introduces the Component Model with typed WIT interfaces and granular capabilities, enabling modules written in different languages to compose without manual serialisation. The full standard arrives in 2024 with direct impact on edge functions, plugins, and serverless.