By Marty Cagan · svpg.com · LinkedIn
Marty Cagan's first published article, from a 1986 HP Journal issue, is a technical overview of Hewlett-Packard's internal AI Workstation research program and its first product, a Common Lisp development environment. It argues that symbolic programming and Lisp can run efficiently on inexpensive workstations, making AI tooling practical and useful even for conventional software development. The piece covers the market rationale, the Lisp language features that favor rapid iteration, object-oriented and rule-based extensions, the integrated editing and browsing environment, and example applications such as diagnostics, simulation and natural language. Republished in 2021 as a time capsule, it is valuable for seeing how early expectations about AI, speech recognition and developer productivity played out over the following decades.
01Key takeaways
- Interactive, incremental development shortens the feedback loop compared with edit-compile-link-test cycles.
- Choosing a language or paradigm should weigh development cost, not just runtime efficiency, once hardware gets cheap.
- Encapsulation and inheritance let teams change internal representations without breaking dependent code.
- Tools that let users customize and extend their environment make the product more flexible than fixed interfaces.
- Early predictions about AI timelines were often optimistic; many capabilities took decades to appear.
02Key sections
- Market and purpose
- HP framed its AI Workstation as an enabling platform that serves both AI-specific applications and productivity gains for conventional software developers. It was designed to run on low-cost hardware, with a server model for distributing applications.
- Why Lisp for rapid development
- The article explains Lisp's advantages: incremental development without declarations, automatic memory management, interactive debugging, and symbol manipulation. It also notes why Lisp had not yet gone mainstream, including cost, performance and suitability concerns.
- Object-oriented and rule-based paradigms
- Two higher-level extensions are described: an object-oriented layer built on data abstraction, inheritance and polymorphism, and HP-RL, a frame-based rule language with forward and backward chaining for knowledge-based systems.
- Integrated environment
- An extensible EMACS-based editor, browsers for code, mail and documentation, and interactive tooling for multiple languages made the environment a single place to edit, test, debug and document programs.
- Applications and examples
- Example systems include photolithography and IT diagnostic expert systems, instrument interpretation tools, a flight planner and simulator, a natural-language understanding system, and early AI-assisted software engineering tools.
03From the post
“A partnership dedicated to teaching best practices to product teams and product leaders”
04Frameworks mentioned
Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.