<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj" -->

---
title: Vercel&#x27;s Zero: a programming language designed for AI agents
description: Vercel has announced Zero, an experimental programming language designed for AI agents rather than human developers. Key features include structured JSON error...
canonical: https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Vercel&#x27;s Zero: a programming language designed for AI agents | daily.dev
og:description: Vercel has announced Zero, an experimental programming language designed for AI agents rather than human developers. Key features include structured JSON error...
og:url: https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj
og:image: https://api.daily.dev/og/posts/ZVWQxatNJ.png
og:image:alt: Vercel&#x27;s Zero: a programming language designed for AI agents
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Vercel's Zero: a programming language designed for AI agents

**[Collections](https://daily.dev/sources/collections)** · 3 min read · 2 upvotes · 0 comments

## Summary

Vercel has announced Zero, an experimental programming language designed for AI agents rather than human developers. Key features include structured JSON error output for programmatic parsing, native binaries under 10KB without LLVM, explicit side effect declarations, and static memory management. The announcement has drawn significant skepticism. Critics like ThePrimeagen point out a chicken-and-egg problem: AI agents perform best with languages that have large training corpora, which Zero lacks entirely. Ginger Bill (Odin creator) dismisses designing for LLMs as guesswork. George Hotz raises a deeper concern that agents fundamentally can't handle the judgment-intensive final 20% of real software projects, and that weak feedback loops in large teams will amplify low-quality AI-generated code. The target audience for Zero remains unclear — systems programmers already have Zig and Rust, while AI-focused teams are entrenched in TypeScript.

## Content

## What is Zero?

Vercel recently announced Zero, an experimental systems programming language designed specifically for AI agents rather than human developers. The pitch is straightforward: instead of printing human-readable compiler errors, Zero emits structured JSON error codes so an agent can parse its own bugs and fix them programmatically. It also compiles to native binaries under 10KB without LLVM, enforces explicit side effect declarations, and uses static memory management.

Some of those design choices are genuinely interesting. Structured error output makes sense if your consumer is a program rather than a person. But the announcement has drawn a lot of skepticism, and the criticisms are worth taking seriously.

## The chicken-and-egg problem

The most obvious issue: AI agents are best at languages with massive training data. TypeScript, Python, Rust, Go — these have millions of examples baked into every major model. Zero has essentially none. So the agents you'd want to use Zero are the least equipped to write it.

ThePrimeagen made this point bluntly: why would a brand-new language with no LLM training corpus be better for agents than TypeScript or Rust? He also noted that Zero looks a lot like Zig and Go without adding anything clearly novel, and that it's verbose rather than token-efficient — which matters if you're paying per token or working within context limits.

Ginger Bill, the creator of Odin, put it more colorfully: designing a language for LLMs is like reading horoscopes. You're optimizing for something you don't actually understand.

There's also a market fit question. The developers most likely to adopt a new systems language — people who care about binary size, memory control, and compile speed — aren't particularly interested in optimizing for AI agents. And the teams leaning hardest into AI-assisted development are mostly happy with TypeScript.

## The deeper skepticism

George Hotz has a more fundamental objection that goes beyond Zero specifically. After six months of experimenting with agents in tinygrad and hardware reverse engineering, his conclusion is that agents are useful for prototyping and search but fall well short of actual software engineering. They frontload progress — you get something working fast — but they never achieve polish. The last 20% of a real project, the part that requires judgment and iteration, is where they fall apart.

His concern about organizations is pointed: large teams with weak feedback loops will suffer most. Low performers can generate high volumes of plausible-looking code without the self-correction ability to notice when it's wrong. The result is what he calls "slop" — code that looks fine until it doesn't.

Hotz aligns with LeCun and Gary Marcus in thinking current LLM architectures can't truly program. Real programming agents, if they ever exist, would need something closer to a world model rather than the pattern-matching that RLVR-style training produces.

## So what's actually going on here?

I genuinely don't know how to feel about Zero. Some of the design decisions are defensible — structured error output for agent consumption is a real idea, not just marketing. But the framing of "a programming language for agents" raises more questions than it answers.

Who is the target user? Not systems programmers, who have Zig and Rust. Not AI teams, who are already deep in TypeScript. The agents themselves can't adopt a language — the humans building and prompting them have to.

Maybe Zero is an experiment worth running. But right now it looks more like a concept than a solution to a problem anyone has clearly articulated.

## Similar posts on daily.dev

- [Vercel Zero: A Programming Language For AI Agents \(2026\)](https://daily.dev/posts/vercel-zero-a-programming-language-for-ai-agents-2026--9sl8aydqj) · Alex CloudStar · 0 upvotes · 0 comments
- [Vercel Labs Ships Zero: A Graph-First Language Built So Agents Write the Code](https://daily.dev/posts/vercel-labs-ships-zero-a-graph-first-language-built-so-agents-write-the-code-cm7iaxkvi) · InfoQ · 0 upvotes · 0 comments

---

Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#compiler](https://daily.dev/tags/compiler), [#vercel](https://daily.dev/tags/vercel)

[View this post on daily.dev](https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Vercel's Zero: a programming language designed for AI agents","url":"https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj"},"datePublished":"2026-05-19T02:12:00.436Z","dateModified":"2026-05-25T04:59:22.233Z","description":"Vercel has announced Zero, an experimental programming language designed for AI agents rather than human developers. Key features include structured JSON error...","image":"https://pbs.twimg.com/media/HIlIaIaW8AAsFVM.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HIlIaIaW8AAsFVM.jpg","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/vercel-s-zero-a-programming-language-designed-for-ai-agents-zvwqxatnj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,compiler,vercel","timeRequired":"PT3M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Vercel's Zero: a programming language designed for AI agents"}]}
```

