Vercel Releases Eve: A File-Based Framework for Building Modular AI Agents

The open-source framework lets you organize agents as directories of files, making it easier to scale and customize AI workflows.

If you build or manage AI agents, Vercel’s new open-source framework, Eve, offers a fresh way to structure them. Instead of treating an agent as a monolithic block of code, Eve maps each agent to a directory of files—each representing a specific capability.

This modular approach simplifies development, debugging, and scaling, according to MarkTechPost.

Why File-Based Agents Matter

Eve’s file-centric design means you can:

1. Isolate functionality – Each agent capability (e.g., text generation, API calls) lives in its own file, making debugging and updates cleaner.
2. Version control with ease – Since agents are just directories, Git workflows apply naturally.
3. Reuse components – Swap or share individual capabilities between agents without rewriting entire systems.

This approach mirrors trends in software development, where microservices and modular architectures dominate.

Vercel isn’t alone in this space—Microsoft’s Orchard and AWS’s Strands also embrace modular agent design.

But Eve stands out by leveraging Vercel’s expertise in developer tooling.

How Eve Works

Here’s what’s inside an Eve agent directory:

• skills/ – Contains files for discrete abilities, like generating text or calling APIs.
• memory/ – Handles short-term and long-term data storage.
• config/ – Defines how the agent behaves in different contexts.

Because everything is file-based, you can edit, copy, or version each piece independently. For example, swapping a text-generation skill for a different model means replacing one file—no need to refactor the entire agent.

Where This Fits in the AI Agent Ecosystem

Eve is part of a broader shift toward open, composable frameworks. If you’ve used platforms like Podify or Synthflow, you know how powerful—but sometimes rigid—prepackaged agents can be. Eve gives you more control, making it a strong fit for developers who need to tailor agents to niche workflows.

For deeper comparisons, see our breakdown of open-source AI agent frameworks.

What’s Next

Vercel hasn’t announced commercial plans for Eve, but its open-source release suggests a play for developer mindshare. If the model gains traction, expect integrations with Vercel’s existing cloud platform—similar to how Deploy AI and Flamehaven bundle hosting with agent tools.

For now, Eve is worth exploring if you:

• Prefer file-based workflows (like config-as-code setups).
• Need to debug or modify agent behavior at a granular level.
• Want to avoid vendor lock-in for core agent logic.

If you’re building AI agents, check out the latest frameworks to see which approach fits your stack.