Thoughtware

About

Why these notes exist, and what they are trying to make visible.

Thoughtware began as a naming problem. The author could describe models, prompts, embeddings, retrieval, tools, agents, memory, and evaluation, and still could not say what kind of system was taking shape. Decisions had become real before their architecture had become visible.

This site collects that vocabulary and the ideas around it: language as the abstraction above thinking, the evolution from deterministic software toward systems that carry judgment, cognitive units and agents with contracts, memory as architecture, evaluation as engineering, and the posture through which a system earns trust over time.

What you will find

  • A concept web of notes drawn from the books, readable in any order
  • Architecture pages for cognitive units, agents, deterministic code, memory, evaluation, and lifecycle
  • A shelf of books and writing that develops individual ideas
  • A glossary of terms with links into the web
  • Studio, an instrument that proposes a Thoughtware framing for a case you describe

Nothing here is a product pitch. The notes are unfinished on purpose. Better formulations are welcome, the distinctions matter more than the particular words.

The books

  • Introduction to Thoughtware, the theory through construction
  • The Cognitive Unit, the named decision and what naming buys
  • Thoughtware: Designing in the Intelligence Age, twelve principles for products when intelligence participates in the work

Start wherever the pressure is: the naming problem, language, or framing from decisions.