Vercel Deepens Its AI Push With New Tools That Build Reliability and Intelligence Into Software

Vercel, the cloud platform for front-end developers, is expanding deeper into AI. The company has rolled out a series of tools designed to help developers build faster, more reliable, and more secure AI applications. Among them: enhancements to the company’s agent that now empower it to investigate anomaly alerts, a new open-source kit that bakes reliability into software from the start, and a marketplace debut for AI agents and services.

Reducing Incident Alert Fatigue

Developers can now leverage Vercel’s agent to monitor their applications for any issues and recommend fixes automatically. By using AI, developers are spared the hours spent diagnosing production issues and manually analyzing metrics across multiple dashboards and third-party tools. Instead, the Vercel Agent can execute anomaly alerts, sending notifications when it proactively detects unexpected or unusual behavior in the app, infrastructure, or data.

First introduced in June, Vercel Agent was an AI assistant integrated into Vercel’s dashboard that tracked app performance and security data. At the time, the company stated that the bot focuses on “Observability, summarizing anomalies, identifying likely causes, and recommending specific actions.” It quickly evolved into a multi-purpose bot after adding code reviews to its capabilities in September. Now, it has another skill: Investigating. But more than that, Vercel is making its agent more widely available, moving it from private to public beta.

It’s worth noting that Vercel Agent is different from the company’s v0 agent. The former is described as a background agent that aids tasks on the Vercel platform, while the latter is a synchronous agent for live vibe coding.

The problem Vercel wants to solve: Alert fatigue. According to the company, “traditional monitoring tools flood teams with notifications based on static thresholds. Teams become desensitized to alerts, and critical incidents slip through while developers waste time investigating false positives.” Using AI, the Vercel Agent acts as a senior engineer, monitoring applications for any unusual activity 24/7/365—are there spikes in billable metrics like function duration, fast data transfers, or increases in so-called 5xx errors?

Developers will no longer need to jump between apps and dashboards to track down the issue, spend countless hours investigating, or worry about knowledge gaps. Vercel Agent promises to handle all of this, along with any post-incident documentation and post-mortem analyses.

Vercel Agent investigations are now available in public beta, but only for those paying customers on the Pro and Enterprise plans with Observability Plus. The company discloses that pricing is usage-based.

A Dev Kit to Build Durable Apps

With Vercel’s new Workflow Development Kit (WDK), developers can create apps that are resilient from the ground up. This open-source TypeScript framework includes an execution model for long-running code, enabling developers to write applications that survive restarts, deploys, and failures—all without managing queues, schedulers, or databases. As with many dev kits, developers need only use two declarative directives in their code.

The WDK works with any framework—starting with Next.js and Nitro, but Svelte, Nuxt, Hono, and Bun are planned in the future—, platform, and runtime and is designed for systems that are both “intelligent and dependable.” As Vercel notes in a blog post, “AI agents that reason across long contexts need to pause between API calls. RAG pipelines that ingest and embed data over time need to survive crashes during multi-hour processing. Commerce systems that wait for user confirmation need to pause for days without consuming resources.” Vercel’s WDK removes a complex process from the developer’s plate, freeing them up to focus more on the logic instead of also on the app’s infrastructure.

To demonstrate how the Workflow Development Kit works, Vercel has created pre-built template use cases for developers to explore: A story generator Slack Bot, a RAG agent, a birthday card generator, a flight booking app, and a natural language image search tool.

Find Agents and Services on the Vercel Marketplace

The final announcement focuses on improvements to the Vercel Marketplace. Developers can now discover AI agents and services to improve their project workflows.

In other words, as developers build out their applications and they want to infuse AI to tackle some of the mundane, but required tasks to develop, deploy, and scale their work, they can search on Vercel’s Marketplace for a solution, similar to finding a plug-in on WordPress.

First introduced in August 2024, the Vercel Marketplace has gradually evolved to be a destination for natively integrated offerings.

Some of the first agents on the Vercel Marketplace include CodeRabbit, an AI platform that assists engineering teams with automated code reviews and change analysis; Corridor, which focuses on code security and compliance guardrails; and Sourcery, which reviews every project update to quickly detect and eliminate bugs or vulnerabilities.

As for services—the tools Vercel says help developers create, customize, monitor, and scale agents—the first batch rolling out includes:

• Braintrust: This trains, tests, and manages custom agents with an evaluation framework
• Kubiks: This orchestrates multi-step AI workflows and fixes issues proactively, going beyond observability to deliver automated remediation
• Autonoma: This uses AI to automate regression testing to catch bugs before production and requires no lines of code to set up
• Chatbase: This builds, analyzes, and optimizes customer-facing agents
• Kernel: This hosts and scales browser automation infrastructure in the cloud
• Browser Use: This allows AI agents to control the browser to automate web tasks using natural language commands

All of these services can be plugged into a developer’s CI/CD, observability, or automation workflows on Vercel.

“Typically, integrating AI services means managing separate dashboards, billing systems, and authentication flows for each tool,” Vercel’s Chief Product Officer, Tom Occhino, and Product Manager, Hedi Zandi, write in a blog post. “A team using three different AI services might waste hours wiring up each integration before writing a single line of application code.”

“You can now add AI-powered workflows to your projects through native Vercel integrations with unified billing, observability, and installation flows built into the platform,” they add.

Featured Image: An AI-generated image depicting a developer coding on a computer. Credit: Adobe Firefly