A Language For Agents

2026-02-09 · Source: Armin Ronacher's Thoughts and Writings · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Intermediate, long

Summary

The article, written on February 9, 2026, explores the increasing likelihood of new programming languages emerging due to the rise of agentic engineering. It argues that while existing codebases are vast, agents can perform effectively with new languages if they offer a strong value proposition and are designed with LLM training in mind. The author notes that the cost of coding is decreasing, making ecosystem breadth less critical, and provides an example of an Ethernet driver being reimplemented in JavaScript by an agent. The piece details specific language features that agents prefer, such as explicit context, structured syntax (braces over whitespace), explicit flow context, and result-based error handling, while also highlighting features agents struggle with, including macros, re-exports, aliasing, flaky tests, and multiple failure conditions. The author anticipates a surge in new languages, driven by the ability to measure agent performance with different language designs.

Key takeaway

For AI Scientists and Research Scientists considering new language development or evaluating existing ones for agentic workflows, prioritize designs that offer explicit context, structured syntax, and result-based error handling. Your agents will perform better with languages that minimize ambiguity, support local reasoning, and avoid features like macros or implicit aliasing, ultimately leading to more efficient and reliable agent-driven code generation and maintenance.

Key insights

New programming languages designed for AI agents will emerge as coding costs decrease and agentic engineering grows.

Principles

• Agent performance is measurable through task iterations and file changes.
• Explicit context and structured syntax aid agent comprehension.
• Language design should prioritize agent-friendliness over human brevity.

Method

Design new programming languages by prioritizing explicit context, structured syntax, result-based error handling, and local reasoning, while avoiding macros, re-exports, and aliasing, to optimize for agent performance and measurability.

In practice

• Use TypeScript for agent-assisted JavaScript development.
• Port libraries with agents instead of complex native bindings.
• Implement effect markers for explicit function dependencies.

Topics

• Agentic Engineering
• Programming Language Design
• AI Agent Development
• Code Generation
• Software Tooling

Best for: AI Scientist, Research Scientist, AI Engineer, Software Engineer, AI Architect

Related on AIssential

• JavaScript Was Designed for Humans. What If We Built a Language for AI? — Programming languages designed for humans hinder AI code generation; an AI-native layer is needed for explicit machine reasoning.
• I Kept Hearing About AI Agents Everywhere — So I Finally Sat Down and Learned What They Are — AI agents are goal-pursuing systems that reason, use tools, and operate in loops, fundamentally changing software interaction.
• A quick chat with Ines & Matt — Agentic systems can generate "software 2.0" by encoding behavior in data examples, not just explicit code.
• Context Is the New Code — Patrick Debois, Tessl — Effective AI agent performance hinges on meticulously engineered context, necessitating a dedicated Context Development Life Cycle.
• Faster Code, Slower Understanding: A Survival Note from a Year of Agent-First Engineering — AI shifts the engineering bottleneck from code production to comprehension and maintenance, demanding new optimization strategies.
• Q&A: What is agentic AI today, and what do we want it to be? — Agentic AI extends generative models with tools to perform actions, facing data scarcity and human oversight risks.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Armin Ronacher's Thoughts and Writings.