June 2026 Summaries

Vercel offers a streamlined platform for deploying server containers, allowing developers to easily build, store, deploy, and autoscale applications using a simple Dockerfile.vercel configuration. By supporting various frameworks such as Go, Rails, Spring Boot, and more, Vercel facilitates deployment by ensuring that applications listen on a designated port and communicate via HTTP. This service eliminates the need for local daemons, registries, or clusters, and it provides dynamic autoscaling based on traffic, charging users only for the CPU time utilized by their code. Vercel's architecture emphasizes rapid startup through optimized boot images and supports observability with integrated logs, traces, and metrics. The platform ensures that each deployment is stateless, enabling seamless scalability and integration with backing services for persistent state. With its focus on minimal configuration and efficient scaling, Vercel seeks to unify backend and frontend deployments under one cohesive system.

Vercel Services offers a unified platform for developers to run multiple frameworks within a single Vercel Project, integrating frontend and backend components seamlessly. This innovation enables atomic deployments, where all services deploy or roll back together, and facilitates internal service communication without needing public internet routing. The platform supports various frameworks with automatic configuration and provisioning, allowing services to communicate internally using service bindings. Vercel's infrastructure includes features like Fluid compute, autoscaling, and Active CPU pricing, optimizing resource use and costs. It provides a comprehensive suite for backend needs, including secure connections, real-time WebSocket support, and integration with external services using short-lived credentials. Additionally, Vercel offers database provisioning, background job processing, and secure compute options, all running under a unified, zero-configuration setup that supports both the frontend and backend in one cohesive environment.

At Vercel Ship 26 in New York, Vercel and Shopify announced a collaborative project to rebuild Hydrogen, a headless storefront solution, to enhance its portability and openness for web developers. The new version will be open-source and runtime agnostic, allowing developers to use various frameworks like Svelte, Nuxt, or Next.js, or even custom frameworks. The architecture consists of three layers: core, client, and server, designed to streamline development by centralizing previously dispersed code, simplifying cart state management, and offering full-stack capabilities without proprietary lock-in. Hydrogen's integration with popular frameworks like Next.js promises seamless handling of static and dynamic content and offers developers guidance through documentation and templates. Vercel aims to incorporate insights from its vercel.shop template into Hydrogen, further enhancing the developer experience in building scalable, high-performance storefronts. The initiative underscores Vercel's commitment to fostering an open web development environment, encouraging community involvement via GitHub to shape its future iterations.

AI Gateway has expanded its support to include audio and voice capabilities, enabling real-time voice interactions, text-to-speech, and speech-to-text functionalities alongside existing text, image, and video modalities. This update, available in AI SDK 7, allows developers to integrate voice features into their applications using models from OpenAI and xAI, with the same routing, observability, and budget controls used for other models. The real-time voice feature allows applications to hold natural conversations by immediately responding to user input, enhancing the user experience for voice assistants and customer support tools. Text-to-speech and speech-to-text features facilitate the creation of audio content from text and vice versa, supporting tasks like voiceovers and transcription. Users can test audio models directly in their browsers without writing code, and audio calls are managed with the same API key and provider controls as other AI Gateway services, streamlining the integration of speech capabilities into existing applications.

Vercel's Open Source Program is dedicated to supporting and nurturing the development of open-source projects by offering resources and assistance to project maintainers. The program, which welcomes new cohorts four times a year, provides maintainers with $3,600 in Vercel platform credits, third-party service credits, and guidance from both Vercel and the community. The Spring 2026 cohort includes diverse projects such as ASCII Studio, a tool for converting video frames into ASCII art; Sync UI, a React UI library; xmcp, a TypeScript framework for MCP servers; and Tirith, a terminal security tool, among others. Each project addresses different technological needs, from enhancing UI components to improving security protocols and facilitating learning platforms, showcasing the breadth and innovation inherent in the open-source community.

AI SDK 7, a TypeScript SDK with over 16 million weekly downloads, enhances the development of AI applications by providing robust tools for building, running, integrating, observing, and extending AI agents across various model providers. It introduces new capabilities such as reasoning control standardization, runtime context, provider file and skill uploads, and MCP Apps for richer agent experiences. The update supports the development of well-behaved agents through fine-grained control over model reasoning, tool context, and file handling, while also offering enhanced observability with telemetry and Node.js tracing channel integration. AI SDK 7 also includes features like real-time voice support, video generation, and a terminal UI for quick agent testing, promoting a more flexible and efficient workflow for developers. Key improvements in agent execution include tool approvals, durable WorkflowAgent support, timeout configurations, and sandbox support, along with experimental harness abstractions for integrating various agent runtimes. This release is a result of collaborative efforts from Vercel's core team and the wider community, aiming to facilitate scalable, autonomous AI agent development with improved performance metrics and lifecycle event tracking.

Coding agents can quickly produce user interfaces by replicating existing patterns and styles, but they struggle to understand the reasoning behind those patterns, which is often found in design reviews and discussions that are not part of the codebase. Vercel addresses this by implementing a system called product-design, which encapsulates product decisions as code, making them available to coding agents. This system comprises an agent skill providing context for product decisions, linters enforcing clear rules, and a review loop that compiles evidence from various sources to update guidelines. The product-design skill is integrated into the repository, organizing decisions, references, and examples to guide agents. The process emphasizes clear decision-making, separating facts from assumptions, and ensuring that changes align with user tasks and product standards. Linters offer fast feedback by enforcing rules, while human oversight is maintained through regular evidence reviews and updates, ensuring that the guidance remains current and effective. This structured approach allows teams to build a similar system tailored to their own standards, enabling agents to make informed decisions while freeing human reviewers to focus on complex judgment calls.

Vercel Flags is an integrated feature flagging solution that allows for seamless deployment and management of feature updates, infrastructure changes, and more within the Vercel platform. By enabling server-side evaluation of flags, Vercel Flags ensures minimal impact on page performance, offering a more efficient alternative to client-side flag evaluation. The platform provides a comprehensive dashboard for managing flags alongside deployments, and the Flags SDK supports various frameworks, including Next.js and SvelteKit. Vercel Flags allows developers to separate code deployment from feature releases, facilitating controlled rollouts and quick rollbacks without altering source files or redeploying. The solution is designed to handle complex scenarios, such as progressive rollouts and infrastructure migrations, with ease and reliability, and is available across all Vercel plans.

Agents are versatile tools capable of handling diverse tasks such as answering support tickets and writing code, requiring three fundamental capabilities: connecting to models, executing complex workflows, and integrating with data and user platforms. The Agent Stack provides a comprehensive suite of tools to build production-grade agents, connecting them to various models via the AI SDK and AI Gateway, which unify and streamline model interactions. Workflow SDK enhances the durability and efficiency of agent tasks, while Vercel Sandbox provides secure, isolated environments for executing unreviewed code. Vercel Connect facilitates secure access to external systems with short-lived tokens, and Chat SDK enables integration across multiple communication platforms. The framework eve encapsulates these components, offering a structured, open-source approach to agent development, allowing developers to focus on writing agents without dealing with underlying complexities.

Eve is an open-source agent framework designed to simplify the process of building, running, and scaling agents by providing built-in production features such as durable execution, sandboxed compute, and human-in-the-loop approvals. It allows developers to focus on defining an agent's purpose without the need to assemble underlying components, similar to how Next.js transformed web development. An agent in eve is structured as a directory of files, each representing a component like the model, instructions, tools, and channels, allowing for modular and scalable development. The framework includes features for secure connections, agent sandboxing, tracing, and evals to ensure reliability and ease of maintenance. It supports integration with various platforms such as Slack, GitHub, and Vercel, enabling agents to operate across multiple channels and automate tasks like scheduling reports. Eve's adoption at Vercel has significantly increased agent-triggered deployments, reflecting its potential for productivity improvements and streamlined workflow management.

Vercel Connect, currently in Public Beta, offers a solution to the security risks associated with long-lived provider tokens by introducing runtime credential exchange. This system enables agents to access external systems on behalf of users and teams without storing permanent provider secrets in environment variables. Instead, short-lived, task-specific tokens are requested at runtime, ensuring that access is precisely scoped and temporary. This approach mitigates the danger of token leaks, as the tokens are only valid for specific tasks and are refreshed automatically. Vercel Connect supports a variety of connectors, including Slack, GitHub, and Linear, and integrates with existing frameworks such as eve, allowing for seamless connection and management of provider credentials. Its architecture emphasizes security by eliminating stored secrets, limiting access to the necessary scope, and enabling token revocation, depending on provider support. This setup is particularly beneficial for applications needing delegated runtime credentials and user authorization without compromising security.

Vercel Ship 2026, held in London, showcased Vercel's innovative approach to agentic infrastructure, emphasizing the integration of artificial intelligence and automation in software deployment. CEO Guillermo Rauch highlighted Vercel's commitment to creating a platform where coding agents can deploy, build, and manage software autonomously. The event introduced key components like the Agent Stack, Vercel Connect, and the open-source framework eve, which streamlines the development and deployment of production agents. Vercel also unveiled its support for full-stack applications, including backend frameworks and databases, as well as Vercel Services, enabling seamless microservices communication. The conference featured discussions on enterprise applications, security, and the future of AI, with insights from industry leaders and Vercel's own experiences in building scalable agentic solutions. The event concluded with engaging talks and panels on the practical implementation of agents in production and the evolving role of founders in the AI-driven landscape.

Vercel for Enterprise Apps and Agents is a platform designed to enable companies to safely deploy AI-powered applications and agents while maintaining strict access and security controls. The platform addresses challenges related to ownership, access, and security by incorporating components such as Vercel Passport, which secures internal apps and agents through integration with identity providers like Okta and Auth0, and Vercel Connect, which provides secure, short-lived access credentials for various systems. Additionally, Enterprise Managed Users offers lifecycle management of user accounts, ensuring proper access control and auditability. The platform also introduces v0, an AI app builder that integrates with Snowflake, allowing users to build data apps directly on their data warehouse. For enterprises needing more control over their infrastructure, the platform offers a "bring your own cloud" option on AWS, enabling deployment within a company's own AWS account. Overall, the platform is designed to facilitate rapid and secure app development by embedding security measures into the deployment process, allowing for innovation without compromising data security.

Okara is an AI CMO platform that helps founders and small teams manage their marketing strategies without needing to hire additional staff. By utilizing a multi-provider AI stack on Vercel, Okara processes 4 billion tokens daily, allowing it to create and execute marketing plans through specialized sub-agents focusing on SEO, content, and social media. This approach enables businesses to enhance their brand voice and distribution efficiently across multiple channels, ensuring visibility and awareness without the high costs typically associated with traditional marketing agencies. The platform's integration with Vercel AI Gateway and Sandboxes streamlines the management of various AI models and automates technical solutions, enabling rapid deployment and scaling for over 120,000 businesses. Okara's infrastructure allows it to support a large client base with a small team, continually expanding its capabilities to serve larger teams while keeping infrastructure demands minimal.

Fern, a platform that assists companies in delivering developer documentation and SDKs, has achieved significant performance improvements by migrating its infrastructure to Vercel. The migration involved moving 65% of its platform from the Pages Router to the App Router in just seven days, which was a more efficient approach than a potentially lengthy engineering project. This transition has resulted in a 3x faster time to first byte and an 80% reduction in page load times, enhancing the user experience for over 6 million page views from more than 1 million unique visitors per month. Fern now benefits from the ability to deploy multiple times a day, with deployment times reduced to five minutes, further optimizing its operations for clients like Webflow and ElevenLabs.

The Weather Company serves over 350 million monthly active users worldwide by delivering weather forecasts and content, supported by approximately 130 meteorologists monitoring global weather systems. By reconstructing its web serving stack and content management system on Vercel and v0, the company significantly reduced the time required to transform a design into a live website, accelerating this process from days to mere hours. This technological upgrade resulted in an 80% increase in velocity from design to live webpage, effectively turning work that previously took years into tasks completed within months.

Code and Theory, a digital-first creative and technology agency working with brands such as Microsoft, Amazon, and NBC, has significantly improved its prototyping process by adopting a prompt-to-code workflow. This innovative approach eliminates the traditional use of wireframes and requirement documents, allowing product, engineering, and go-to-market teams to transform client briefs into functional prototypes in real-time. As a result, the agency has reduced time-to-prototype by 75% and cut deployment timelines by over 50%, enhancing efficiency and responsiveness in its project delivery.

In May 2026, AI Gateway reported significant shifts in AI token routing and spending patterns, revealing a 20% month-over-month increase in total tokens and a 43% increase in spending. DeepSeek's introduction of its V4 models marked a major change, capturing 17% of token volume while maintaining a low cost share, highlighting their effectiveness and affordability for high-volume tasks. Anthropic continued to dominate high-stakes use cases, holding a significant portion of the spend despite increased competition from lower-cost models like DeepSeek. The report underscores a growing trend towards strategic model routing, with teams optimizing their budgets by employing cheaper models for less critical tasks and reserving more expensive, high-performance models for areas where quality is paramount. This approach reflects a heightened pricing sensitivity among developers, who are increasingly focused on maximizing the efficiency of their AI investments. Additionally, the report notes a reluctance to adopt new, more expensive models without clear return on investment, emphasizing a pragmatic approach to AI deployment in production environments.