September 2025 Summaries

Vercel has achieved a significant milestone with a Series F funding round that values the company at $9.3 billion, accompanied by a $300 million investment co-led by Accel and GIC, and a tender offer for early stakeholders. Celebrating its 10th anniversary, Vercel is experiencing its most transformative year, marked by a shift from building web pages to developing AI-powered agents, reflecting a broader evolution from pixels to tokens. This transformation is underscored by the launch of the AI Cloud, designed to support AI applications, and the development of an AI SDK that offers access to over 60 models, resulting in over 3 million weekly downloads. Vercel's commitment to open source and user-focused innovation is exemplified by its open-sourcing efforts, the widespread adoption of Next.js, and the beta release of Vercel Agent for AI-driven code reviews. The upcoming release of the v0 iOS app, currently in public beta, aims to enhance productivity by enabling development outside traditional work hours, demonstrating Vercel's vision for a dynamic web where imagination seamlessly transitions into creation.

Claude Sonnet 4.5, now available on Vercel AI Gateway with full support in AI SDK, offers enhanced capabilities for developers by improving reasoning and coding tasks, particularly in building and linting Next.js applications. This version exhibits better design sensibility, code quality, and agentic coding workflows, which include a higher success rate in Next.js builds, improved linting, and better handling of framework-specific functionalities. In conjunction with the Vercel AI Cloud, it enables a new class of developer workflows where AI can autonomously plan, execute, and ship changes. The new Coding Agent Platform template leverages Claude Sonnet 4.5 to automate task execution within GitHub repositories, providing a multi-agent system that integrates tools like Claude Code and OpenAI Codex CLI. This platform deploys tasks in isolated sandboxes, ensuring secure and efficient execution with real-time progress tracking. Additionally, the Coding Agent Platform template is open-sourced, allowing for community-driven development and extension.

Vercel's introduction of request collapsing in the CDN aims to optimize Incremental Static Regeneration (ISR) in Next.js applications by minimizing redundant function invocations during cache expiration scenarios, known as "cache stampedes." When numerous users simultaneously request an expired route, request collapsing ensures only one function invocation per region occurs, while others wait for the cached response, thus conserving compute resources and preventing backend overload. The system automatically determines cacheability through framework integration, requiring no manual configuration. It employs a two-level distributed locking mechanism—at the node and regional levels—to coordinate requests efficiently, preventing a thundering herd problem. The process utilizes double-checked locking, checking the cache both before and after acquiring locks to avoid duplicate regenerations. Additionally, the system is resilient to failures, using timeouts to handle function errors and timeouts gracefully, ensuring continuous responses even under high traffic. In production, this feature significantly reduces unnecessary function calls and is automatically enabled for all Vercel projects using ISR.

A financial institution experienced a spike in suspicious bot traffic, initially suspected to be malicious, but later identified as verified Google bots executing strange search queries unrelated to their business. Investigation revealed that their previous infrastructure had been compromised years ago through SEO poisoning, where attackers manipulated search engines to index malicious content. Although the institution had since moved to a new platform, Google's index still contained these poisoned URLs, causing its bots to re-crawl the old, corrupted pages. By using BotID for precise bot identification, the institution avoided mistakenly blocking Google's legitimate crawlers and instead focused on updating their robots.txt and using Google Search Console to remove the outdated, harmful pages. This case highlights the importance of accurate bot identification over immediate blocking, as understanding the source of bot traffic can distinguish between active threats and residual effects of past vulnerabilities.

A recent optimization to a global routing service has significantly enhanced performance by implementing a Bloom filter to replace a slow JSON parsing operation, resulting in a 15% reduction in memory usage, a 10% improvement in time-to-first-byte (TTFB) for requests above the 75th percentile, and faster routing speeds for websites with numerous static paths. The previous method, which involved generating and parsing a large JSON file to check if a requested path existed, particularly affected websites with massive path lists, such as e-commerce platforms and documentation sites, causing delays of up to 250 milliseconds. The introduction of Bloom filters, a probabilistic data structure, has reduced path lookup times to nearly zero, as Bloom filters are smaller and faster than storing the entire path list, and they guarantee no false negatives, only occasionally triggering extra storage requests due to false positives. This transition required implementing matching Bloom filter algorithms across different services and resulted in a 200x to 100x speed improvement over the previous approach, reducing heap size and memory usage by 15%, and consequently improving performance across the entire routing service.

Vibe coding, a concept introduced by Andrej Karpathy in 2025, is transforming the landscape of software development by allowing users to create applications and workflows using AI-driven tools with minimal technical expertise. This approach, characterized by the notion of "what you say is what you get," enables both developers and non-developers to generate projects quickly, as seen with tools like v0, Lovable, and Replit. Large companies such as Amazon and Google have already embraced vibe coding, significantly reducing the time and resources needed for project completion. The rise of vibe coding has redefined the role of developers, shifting the focus from team-based productivity to individual capability, with a growing number of non-developers using these tools to innovate. However, the accessibility of vibe coding raises security concerns, necessitating the integration of robust security measures into the platforms themselves. As interest in vibe coding continues to grow, businesses face the choice of adapting to this new model or maintaining traditional coding methods for projects where security is paramount.

Vercel addresses the long-standing issue of cold starts in serverless computing, which occur when a serverless function must initialize from scratch, causing delays for users. Traditional serverless platforms often face a trade-off between cost and performance, as they shut down inactive instances to save money but introduce latency when traffic returns. Vercel's innovative approach, powered by Fluid compute, nearly eliminates cold starts by keeping instances warm and ready to handle requests, ensuring 99.37% of requests experience no cold starts. Their solution involves multiple strategies, including scaling to one, predictive scaling, and bytecode caching, which together reduce the frequency and impact of cold starts. Additionally, Vercel implements rolling releases to prevent deployment-induced cold starts by gradually shifting traffic to new deployments. This comprehensive strategy ensures that applications remain responsive and efficient, regardless of scale, allowing developers to maintain optimal performance without sacrificing cost-efficiency.

Model Context Protocol (MCP) is gaining traction as a standard for federating tool calls between agents, offering a microservice architecture that enables tool reuse across AI applications. However, using MCP tools in production can introduce risks such as unexpected changes to tool names, descriptions, and schemas, which can lead to security, cost, and quality issues. The mcp-to-ai-sdk addresses these challenges by generating static AI SDK tool definitions from any MCP server, allowing enterprises to lock tool schemas and descriptions in their codebase and update them explicitly. This approach reduces the risks of prompt injection and unexpected capability introductions while controlling costs and improving latency by selectively loading necessary tools. By vendoring tool definitions, companies can benefit from MCP's discovery capabilities during development while ensuring stability and security in production environments. As AI systems transition from prototypes to production, this method provides a balance between flexibility and reliability, emphasizing the importance of secure and reviewable development practices.

Rox is advancing its revenue operating system by utilizing intelligent AI agents for research, prospecting, and engagement, enabling enterprises to manage and grow revenue more rapidly. From its inception, Rox has relied on Vercel's infrastructure to enhance the speed and scalability of its web applications, ensuring fast and consistent user experiences globally. The use of Vercel's infrastructure allows Rox to prioritize development over infrastructure concerns, with features such as image optimization, serverless caching, and preview environments facilitating efficient deployment and testing. Rox's AI agents perform instantaneous research across numerous prospects, overcoming browser limitations through Vercel Serverless Functions, which batch and stream results efficiently. Additionally, the Vercel AI SDK powers Rox Command, a chat-driven interface that provides conversational AI experiences, accelerating product feedback and deployment cycles. This collaboration with Vercel has resulted in rapid innovation and maintained performance, supporting Rox's mission to transform go-to-market strategies with AI.

Helly Hansen, a 150-year-old Norwegian technical apparel brand, successfully enhanced its digital presence and e-commerce performance by transitioning from an outdated tech stack to the modern solutions offered by Next.js and Vercel. This change addressed critical performance issues that previously hampered their site, especially during peak revenue events like Black Friday and Cyber Monday. By shifting from client-side to server-side rendering, they improved Core Web Vitals, reduced load times, and eliminated major layout shifts, resulting in an 80% year-over-year growth during Black Friday with zero downtime. The migration strategy, which began with optimizing the checkout process, proved beneficial in boosting conversion rates and marketing efficiency. Additionally, the move to a modern, API-first infrastructure increased development velocity and allowed frequent A/B testing, leading to a more personalized customer experience and 21% e-commerce growth. Helly Hansen's partnership with Vercel not only resolved immediate performance challenges but also established a scalable and future-proof foundation for ongoing innovation and competitiveness.

Vercel has expanded its Log Drains feature, now called Vercel Drains, to include the ability to export OpenTelemetry traces, Web Analytics events, and Speed Insights metrics, offering a unified method for streaming observability data from Vercel to existing systems like Datadog, Honeycomb, or Grafana. This enhancement allows for a more comprehensive understanding of application performance by combining logs, traces, analytics, and real-user metrics into a single pipeline, providing consistent and correlated data that enriches logs with traceId and spanId for seamless navigation from logs to distributed traces. Drains can be configured as custom HTTP endpoints for various data types or through direct integrations with vendors, facilitating automatic setup and streaming to preferred platforms, and are available on Vercel's Pro and Enterprise plans with data export billed at $0.50 per GB.

x402 is an open protocol designed to simplify the integration of payments into HTTP requests by using the 402 Payment Required status code, allowing API endpoints to request payment without prior account setup. This protocol addresses the challenges AI agents face when accessing paid external services, which typically require manual registration and billing setup. x402 integrates with Model Context Protocol (MCP) servers and the Vercel AI SDK, facilitating programmatic payments as part of the standard HTTP exchange, making it independent of centralized payment providers and adaptable to various payment networks. Implementations currently involve one-time payments in USDC on the Base blockchain, but the protocol remains agnostic to payment networks, supporting both crypto and non-crypto currencies. The x402 AI Starter template on Vercel showcases its application with various AI developer tools, providing a practical starting point for developers interested in building applications that involve automated payments.

MongoDB Atlas is now accessible via the Vercel Marketplace, allowing developers to manage a fully integrated MongoDB database directly from the Vercel dashboard without the need to navigate multiple platforms or manage separate accounts. This integration simplifies database provisioning by automatically setting up environment variables and consolidating billing within the Vercel account, thereby streamlining the workflow for developers. The partnership between Vercel and MongoDB aims to enhance the developer experience by enabling faster application development and scaling through the AI Cloud, leveraging MongoDB's flexible data model, built-in search capabilities, and robust security features. MongoDB Atlas offers a range of deployment options and supports advanced querying and AI/ML functionalities, making it a powerful addition to Vercel's suite of tools for building high-performance applications. The integration promises to deliver a modern data layer that facilitates the development of web and AI applications, with easy database connectivity through automatic configuration of the MONGODB_URI environment variable upon provisioning.

Vercel is updating its Pro plan to enhance collaboration, infrastructure consumption, and workload evolution by introducing a flexible spending model, free Viewer seats, and self-serve access to Enterprise features like SAML SSO and HIPAA BAAs. The new credit-based usage model starts with $20 in flexible credit for teams, supplemented by dedicated funds for Fast Data Transfer and Edge Requests. The plan includes improved Spend Management with automatic alerts to help prevent unexpected costs. These changes, which may lead to modest bill increases for about 7% of teams, aim to support modern team-building practices by offering scalable solutions and optional features without requiring sales contact or contracts, and are designed to accommodate both small teams and those operating on a global scale.

The text discusses the challenges and solutions related to the implementation of MCP (Machine Communication Protocol) tools when interacting with large language models (LLMs). Initially, many teams created superficial MCP implementations by simply wrapping existing APIs to quickly claim support for the standard. However, this approach proved inefficient for LLMs, which lack the ability to recall past interactions or maintain state across conversations, leading to repeated and inconsistent orchestration of API calls. The text suggests that instead of exposing low-level API operations, tools should be designed around complete user intentions, such as deploying a project end-to-end, to streamline processes and reduce complexity for LLMs. This involves creating comprehensive workflow tools that manage the entire task internally and provide conversational updates, thus improving reliability and efficiency. The article emphasizes testing workflows manually before coding and adjusting tool design based on real user interactions to ensure complex workflows are successful on the first attempt, ultimately leading to better performance and user experience.

On September 8, 2025, a significant supply chain attack compromised 18 popular npm packages, including chalk, debug, and ansi-styles, by injecting malicious code designed to intercept cryptocurrency transactions in browsers. This attack was part of a coordinated effort that extended to DuckDB-related packages after the duckdb_admin account was breached. Although Vercel customers were not affected by the DuckDB incident, the company took immediate action by identifying affected projects, purging build caches, and notifying impacted customers with specific guidance. The attack originated from a phishing campaign targeting npm package maintainers, using a fraudulent domain to harvest credentials under the guise of a two-factor authentication update. Vercel emphasized strengthening supply chain security by enhancing monitoring and improving tools for rapid cache invalidation, while also advising customers to rebuild affected projects, review dependency update practices, and use npm audit and dependency scanning. The incident underscores the importance of defense-in-depth strategies, rapid detection, and response to minimize impact, with contributions from Aikido Security and the npm community playing a crucial role in addressing the compromised packages.

Vercel recently collaborated with the Biome team to enhance the noFloatingPromises lint rule, which aims to identify unhandled Promises that can lead to silent errors in code. To test the robustness of the updated rule, Vercel organized an internal competition to devise complex examples that would challenge the lint rule, emphasizing creativity and collaboration among engineers. The competition highlighted various edge cases, such as using arrays of Promises, Promise-like objects, and JavaScript features like Proxy objects and optional chaining, to showcase how Promises could inadvertently remain unhandled. The Proxy Promise example was deemed the most innovative, demonstrating the complexities of catching such errors in real-world scenarios. This initiative not only improved the lint rule but also reinforced Vercel's commitment to fostering a collaborative culture focused on enhancing developer experiences and supporting open-source projects.

Vercel emphasizes its commitment to open source development by promoting an "Open SDK" strategy that advocates for open, portable, and platform-agnostic tools and frameworks. The company aims to enhance the overall quality of software without forcing developers into lock-in situations, even as they innovate and build on their own platform. This approach is driven by lessons learned from the development of Next.js and the need for tools like the AI SDK to be framework-agnostic. Vercel's strategy includes expanding its principles to various ecosystems, such as Nuxt for Vue and Svelte, and providing versatile tools such as Nitro and Turborepo. By fostering an open environment, Vercel aspires to attract open source creators and contribute to building a better, more accessible web, aligning business growth with quality and mission.