What is Code

2026-05-12 · Source: Martin Fowler · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Advanced, long

Summary

The article "What is Code," published May 12, 2026, redefines code's role in the LLM era. It argues that while LLMs commoditize code as machine instructions, its enduring value is as a conceptual model of the problem domain, building a shared vocabulary. Coding is an act of translation, mapping domain-specific terms (e.g., "customers," "orders") onto technical vocabulary (e.g., "GET," "POST") and creating new constructs within programming languages. The piece emphasizes "bounded contexts" from Domain-Driven Design for local vocabulary discovery and iterative methods like TDD. Programming languages are presented as thinking tools that aid design discovery through their constraints. The article also addresses how LLMs utilize this vocabulary and warns of "cognitive debt" when generated code introduces terms without shared human understanding. It concludes that robust foundational code with clear abstractions and vocabulary is vital, acting as a "harness and context" for LLMs.

Key takeaway

For AI Engineers or Software Architects designing systems with LLM integration, recognize that code's value increasingly lies in its conceptual model and vocabulary, not just generated syntax. Focus your efforts on building robust, well-defined abstractions and a precise, consistent domain language within your codebase. This foundational work will not only improve human understanding but also serve as a critical "harness and context" for LLMs, making their output more reliable and reducing cognitive debt. Prioritize clarity and shared understanding over raw code generation speed.

Key insights

Code's primary value shifts from machine instructions to building and refining a shared conceptual model and vocabulary.

Principles

• Code is a conceptual model, not just machine instructions.
• Shared vocabulary is key for human and LLM communication.
• Programming languages are tools for design discovery.

Method

The article describes an iterative process of building local abstractions and vocabulary through coding, reflection, and continuous feedback, often using techniques like TDD and close collaboration with domain experts.

In practice

• Employ TDD for iterative vocabulary and abstraction discovery.
• Foster collaboration between developers and domain experts.
• Build precise, consistent code vocabulary for LLM integration.

Topics

• Code Semantics
• Large Language Models
• Domain-Driven Design
• Conceptual Models
• Software Architecture
• Cognitive Debt
• Programming Languages

Best for: Machine Learning Engineer, Software Engineer, AI Engineer, AI Architect

Related on AIssential

• Case study: AI-assisted animation let a solo creator produce a 17-minute anime pilot — AI-assisted animation empowers solo creators to produce complex projects previously requiring large teams.
• Ad Creative Is A Technology Problem And Opportunity — Creative adtech is maturing, shifting from productivity to performance and integrated systems, driven by AI.
• Best Sudowrite Alternatives in 2026: 8 AI Writing Tools I Tested on Real Fiction Projects — AI writing tools for fiction offer creative support, but complex pricing models like Sudowrite's can hinder adoption.
• We’ve Reached the Point Where AI Has to Catch AI (The Internet Is Full of Slop) — AI detection tools are crucial for verifying content authenticity amidst widespread AI-generated text.
• Drifting Preference Optimization for One-Step Generative Models — DrPO enables efficient preference finetuning for one-step generative models using black-box rewards and feature-space updates.
• Google AI Studio vs Gemini App: What’s the Difference? — Gemini App is for using AI, while Google AI Studio is for building and prototyping with Gemini models.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Martin Fowler.