Vercel announced several updates to its AI development tools during the Ship AI event. The event featured keynotes and talks on AI workflows, agents, and observability.
Among the releases was a beta version of AI SDK 6, which adds an agent abstraction layer for defining and reusing AI agents within projects. This layer allows developers to specify agent behavior once and apply it to different parts of the application. The SDK also includes tool execution approval, integrating a human process to review and confirm AI actions before proceeding. Type safety extends across supported AI models and user interfaces to ensure data consistency and reduce runtime errors through compile-time checks.
Vercel Marketplace has received updates to make it easier to discover and integrate AI agents and services. We now support installing agents such as CodeRabbit for code assistance, Corridor for workflow automation, and Sourcery for refactoring tasks. AI services available through the marketplace include Autonoma for data processing, Braintrust for testing model output, Browser Use for web interactions, Chatbase for chat data analysis, Descope for identity management, Kernel for task processing, Kubiks for pipeline operations, and Mixedbread for coordinating multiple models. These integrations connect directly within your Vercel project with integrated billing and simplified setup.
Vercel introduced the use of Workflow, an open source library in TypeScript. The tool transforms standard functions into durable workflows by managing retries for failed operations, executing background steps without blocking the main thread, and providing built-in observability to track execution. It operates independently of any specific framework, allowing it to be used in environments such as React, Next.js, or plain Node.js applications. This library handles state persistence and resumption, making it suitable for long-running processes in AI-driven systems.
Vercel Agent, currently in beta, serves as an intelligence component for deployed applications. Perform AI-based code reviews on pull requests and generate patches that are validated in real-world scenarios before being applied. The agent also monitors for anomalies in your production environment, such as unexpected performance degradation, and initiates automated investigations to identify root causes and suggest remediation. Beta access includes promotional credits for use.
The Vercel Python SDK is now in beta, allowing you to deploy Python-based web frameworks to Vercel AI Cloud without manual configuration. It supports FastAPI for API development and Flask for lightweight servers, and automatically handles scaling and routing in a serverless setup. Developers install it via pip and deploy their projects as they would with their JavaScript equivalent.
To support teams in their adoption, Vercel launched An Agent on Every Desk, a program that provides guidance on implementing AI agents. This includes consulting to identify suitable use cases, access to reference templates, and assistance in moving prototypes into production. For go-to-market functionality, open-source agent templates process leads by enriching data from external sources and qualifying prospects based on criteria such as company size and engagement signals. Another open-source template connects a Slack channel to a SQL database, enables natural language queries for business metrics, and generates responses from the query results.
The community’s response to X highlighted enthusiasm for a practical tool. Developer influencer Matt Pocock posted:
The AI SDK is the primary AI library in the TS ecosystem.
Meanwhile, developer Divin Prince said:
I’ve been playing around with vercel workflows and have written some async/wait code that can pause/resume which is much easier than managing the queue yourself.
Overall, responses praised the focus on production-ready features and requested more documentation on marketplace integration. These beta tools are open to community testing, and Vercel is seeking input to shape future stable versions.
