{
  "id": 16208215,
  "title": "Michigan prof creates Python-like Jac language for AI agents",
  "url": "https://urgent.news/2026/10/23/michigan-prof-creates-python-like-jac-language-for-ai-agents-16208215",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-23T15:46:19.000Z",
  "source": {
    "name": "The Register",
    "slug": "the-register",
    "url": "https://www.theregister.com/devops/2026/10/23/michigan-prof-creates-python-like-jac-language-for-ai-agents/5305110"
  },
  "original_language": "en",
  "account": "University of Michigan professor Jason Mars has developed an ambitious programming language called Jac and a runtime called Jaseci. These tools treat AI and AI agents as first-class entities, according to Mars. In a recent talk at the All Things Open conference, Mars demonstrated how Jac, a Python-like language, can integrate TypeScript and Python runtimes and toolchains. It can compile to Python bytecode, JavaScript, and native machine code, and can utilize libraries from both the PyPI and npm ecosystems. Moreover, Jac incorporates the LLVM compiler to leverage Rust-style borrow checking and ownership semantics, enhancing memory safety and performance. The language also allows importing C and C++ libraries through the LLVM integration. Mars introduced the \"eject function,\" which enables reversing the user's Jac code into TypeScript, Python, or C for corporate compliance purposes. One notable limitation of Jac is the absence of native Java support, as Mars stated, \"If you really need Java, Jac isn't for Java.\" Jac aims to address both ecosystem and architectural fragmentation by defining data structures as graphs, allowing agents to process data more efficiently. This approach can help agents save tokens by avoiding the need to extract data from databases, object blobs, JSON strings, or Python objects. Mars conducted a comparison between Jac and LangGraph, an agent orchestration framework, using an AI agent to build a replica of Greptile, an AI code review tool. The Jac agent was able to autonomously construct a sophisticated multi-stage app that could call multiple tools in parallel, while the LangGraph agent produced a linear state machine that could not execute tasks in parallel. Mars concluded that the quality of AI output is determined by the tools provided to the AI, and Jac has begun to challenge other Python agent frameworks in smaller commercial AI-related services.",
  "summary": "Stop burning $165K on AI rewrites; Jac is meant to \"gobble up\" your polyglot stack",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}