<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3" -->

---
title: How we Solved Agent Building — Andrew Qu, Vercel | daily.dev
description: A Vercel engineer recounts a year-long internal experiment building AI agents for the data team, evolving from a single mega-prompt to a multi-agent pipeline,...
canonical: https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: How we Solved Agent Building — Andrew Qu, Vercel | daily.dev
og:description: A Vercel engineer recounts a year-long internal experiment building AI agents for the data team, evolving from a single mega-prompt to a multi-agent pipeline,...
og:url: https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3
og:image: https://api.daily.dev/og/posts/UxaoC0Bx3.png
og:image:alt: How we Solved Agent Building — Andrew Qu, Vercel
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.

# How we Solved Agent Building — Andrew Qu, Vercel

**[AI Engineer](https://daily.dev/sources/aidotengineer)** · 17 min read · 1 upvotes · 0 comments

## Summary

A Vercel engineer recounts a year-long internal experiment building AI agents for the data team, evolving from a single mega-prompt to a multi-agent pipeline, then to a single stateful agent, and finally to a minimal file-system-based agent inspired by Claude Code and Opus 4.5. That shift doubled internal eval scores. The team then generalized the approach into 'skills' and built Eve, a new open-source agent framework (released two weeks prior at a London event) modeled after Next.js's file-system conventions, letting developers declare skills, tools, and channels as files, with tight integration into Vercel's workflows, sandbox, and Connect products.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.youtube.com/watch?v=9dYcwOkpCE8>

## Questions this post answers

### What is Eve, the agent framework Vercel released?

Eve is an open-source agent framework built by Vercel, designed like Next.js but for agents, using file-system conventions where developers create skills, tools, and channels folders instead of writing custom orchestration code. It supports open-source adapters for Postgres, OpenAI's responses API, and Docker, and integrates natively with Vercel Workflows for durability, Sandbox for secure execution, and Vercel Connect for short-lived OAuth tokens. It was released at a London event and is available at eve.dev.

_Developers weighing agent frameworks can track releases like Eve as they choose their stack on daily.dev._

### Why did switching to a minimal file-system-based agent architecture improve results over a multi-agent pipeline?

A minimal file-system agent with just list-file, read-file, and run-bash tools outperformed a hand-built multi-agent pipeline because it let the model use tools it was already well-trained on and explore emergent behavior instead of following a rigid prescriptive workflow. Switching from a chained multi-agent design to a single file-system agent (inspired by Claude Code and Opus 4.5) doubled eval scores at Vercel's internal data agent.

_Teams debating multi-agent versus single-agent designs can follow real-world experience reports like this on daily.dev._

## Similar posts on daily.dev

- [Introducing eve](https://daily.dev/posts/introducing-eve-at3krhctw) · Vercel · 41 upvotes · 3 comments
- [Vercel's Andrew Qu on why agents are a new kind of software](https://daily.dev/posts/vercel-s-andrew-qu-on-why-agents-are-a-new-kind-of-software-suyqdjwun) · Latent Space · 0 upvotes · 0 comments
- [Vercel Introduces Eve, an Open-Source Framework for Building AI Agents](https://daily.dev/posts/vercel-introduces-eve-an-open-source-framework-for-building-ai-agents-ic7xwvqxp) · InfoQ · 2 upvotes · 0 comments
- [Vercel launches eve, an open-source framework that treats agents as directories](https://daily.dev/posts/vercel-launches-eve-an-open-source-framework-that-treats-agents-as-directories-8jfizvhze) · The New Stack · 1 upvotes · 0 comments
- [How to use Vercel eve: The Next.js framework for AI agents](https://daily.dev/posts/how-to-use-vercel-eve-the-next-js-framework-for-ai-agents-8cdmpdnsr) · LogRocket · 6 upvotes · 1 comments

---

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

[View this post on daily.dev](https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3)

```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":"How we Solved Agent Building — Andrew Qu, Vercel","url":"https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3"},"datePublished":"2026-09-14T17:19:19.757Z","dateModified":"2026-09-14T20:43:47.820Z","description":"A Vercel engineer recounts a year-long internal experiment building AI agents for the data team, evolving from a single mega-prompt to a multi-agent pipeline,...","image":"https://i.ytimg.com/vi/9dYcwOkpCE8/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/9dYcwOkpCE8/sddefault.jpg","isAccessibleForFree":true,"articleSection":"AI Engineer","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":"AI Engineer","logo":"https://media.daily.dev/image/upload/s--u5PucxNT--/f_auto/v1724338940/logos/aidotengineer","url":"https://daily.dev/sources/aidotengineer"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,claude-code,vercel","timeRequired":"PT17M","video":{"@type":"VideoObject","name":"How we Solved Agent Building — Andrew Qu, Vercel","description":"A Vercel engineer recounts a year-long internal experiment building AI agents for the data team, evolving from a single mega-prompt to a multi-agent pipeline,...","thumbnailUrl":"https://i.ytimg.com/vi/9dYcwOkpCE8/sddefault.jpg","uploadDate":"2026-09-14T17:19:19.757Z","duration":"PT17M","url":"https://api.daily.dev/r/UxaoC0Bx3","embedUrl":"https://www.youtube.com/embed/9dYcwOkpCE8"}}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"AI Engineer","item":"https://daily.dev/sources/aidotengineer"},{"@type":"ListItem","position":3,"name":"How we Solved Agent Building — Andrew Qu, Vercel"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-we-solved-agent-building-andrew-qu-vercel-uxaoc0bx3#faq","mainEntity":[{"@type":"Question","name":"What is Eve, the agent framework Vercel released?","acceptedAnswer":{"@type":"Answer","text":"Eve is an open-source agent framework built by Vercel, designed like Next.js but for agents, using file-system conventions where developers create skills, tools, and channels folders instead of writing custom orchestration code. It supports open-source adapters for Postgres, OpenAI's responses API, and Docker, and integrates natively with Vercel Workflows for durability, Sandbox for secure execution, and Vercel Connect for short-lived OAuth tokens. It was released at a London event and is available at eve.dev. Developers weighing agent frameworks can track releases like Eve as they choose their stack on daily.dev."}},{"@type":"Question","name":"Why did switching to a minimal file-system-based agent architecture improve results over a multi-agent pipeline?","acceptedAnswer":{"@type":"Answer","text":"A minimal file-system agent with just list-file, read-file, and run-bash tools outperformed a hand-built multi-agent pipeline because it let the model use tools it was already well-trained on and explore emergent behavior instead of following a rigid prescriptive workflow. Switching from a chained multi-agent design to a single file-system agent (inspired by Claude Code and Opus 4.5) doubled eval scores at Vercel's internal data agent. Teams debating multi-agent versus single-agent designs can follow real-world experience reports like this on daily.dev."}}]}
```

