"It runs on the device, so it's private" is the marketing line. It's also a half-truth that has c...

An agent is only as reliable as the parser that reads its output. Chapter 3.1 covered designing t...

Chapter 1.3 established the bandwidth ceiling as the binding constraint on LLM decode: 136.5 GB/s...

Chapter 1.2 laid out the quantization recipes Intel NPU supports: INT8-sym, INT4-sym group-128 or...

This book is about a narrow, awkward, increasingly important corner of applied AI: building agent...

These are the primary sources for the technical claims in the book. Where multiple sources existe...

The book uses vocabulary from three communities that don't always agree on terms: Intel NPU hardw...

Chapter 2 closes here. We have a model that fits, weights we can stream, KV state we can manage, ...

The distinction between KV cache (what you keep in memory) and KV cache bandwidth (what you strea...

The agent's state — what it remembers from past steps and what it uses to make the next decision ...