Agent Product Design System

7/21/2026

This post introduces the expansion of Vercel Agent, detailing its new capabilities for investigating production issues, answering project questions, and taking approved actions directly from the dashboard. It highlights the agent's role as an autonomous first responder to production changes, investigating logs, metrics, and deployments to find root causes and propose fixes. The post elaborates on specific use cases such as reviewing pull requests, tracing cost increases, fixing broken builds, and checking deployment readiness. Crucially, it introduces a new security model for agents, emphasizing their distinct identity (`vercel-agent`), read-only default mode, and the 'plan-to-permission' prompt model for granting scoped, short-lived access for specific tasks. The integration of Vercel Sandbox for running generated code in an isolated environment is also detailed. The concept of 'anti-fragile infrastructure' is presented as the foundation for trusting autonomous systems, where immutable deployments and sandboxing contain mistakes.

6/25/2026

This post details the architecture and implementation of Vercel's 'product-design' skill, a system designed to provide coding agents with product design context. It outlines the repository structure for the skill, including `AGENTS.md`, `SKILL.md`, `references/` (e.g., `product-judgment.md`, `interface-quality.md`, `copy.md`), and `exemplars/`. The post elaborates on the skill's request routing modes (shape, implement, review, copy, harden) and the decision authority hierarchy. It also defines a detailed workflow for applying the skill, from setting scope and mode to loading context, modeling decisions, mapping surfaces, loading routed references, implementing, and verifying the output. The core contribution is the formalization of a system that bridges the gap between code and product design rationale for AI agents.

5/27/2026

This post details how Conductor migrated its parallel coding agent system from local development environments to the cloud using Vercel Sandboxes. It highlights the technical challenge of local hardware limitations and how Vercel Sandboxes provided a scalable, reliable, and fast remote execution layer. The post emphasizes the model-agnostic nature of Conductor's solution and the seamless user experience achieved by abstracting away the underlying infrastructure.

4/9/2026

This post introduces the concept of 'Agentic Infrastructure' as the next evolution of software infrastructure driven by LLMs and coding agents. It outlines three key areas: 1) Infrastructure for coding agents to deploy to, emphasizing programmatic and deterministic deployment surfaces like immutable deployments, preview URLs, and instant rollbacks. 2) Infrastructure for building and running agents, detailing Vercel's unified platform for AI primitives including AI SDK, Chat SDK, AI Gateway, Fluid compute, Workflows, Queues, Sandbox, and Observability. 3) Infrastructure that is itself agentic, where the platform can autonomously respond to production anomalies by investigating data, inspecting code, and performing root-cause analysis, moving towards proactive systems that act on behalf of developers.

3/30/2026

This post introduces the concept of 'leveraging vs. relying' on AI agents, emphasizing the critical need for engineers to maintain judgment and ownership over agent-generated code. It highlights the dangers of false confidence from passing CI and the gap between code correctness and production safety. The post advocates for building a closed-loop system with 'self-driving deployments' (incremental rollouts with automatic rollback), 'continuous validation' (ongoing load and chaos testing), and 'executable guardrails' (operational knowledge encoded as runnable tools). Vercel's investments in this area include stronger guardrails, stricter static checks, production-mirroring E2E testing, read-only agents for invariant verification, and metrics to surface increasing risk. The core message is to leverage agents but own the risk, with engineers being responsible for understanding the impact of code on production.

3/3/2026

Introduces the 'Slack agent skill' and 'Slack Agent Template' which automate the process of building and deploying Slack agents. This includes project setup, Slack app creation, environment configuration, local testing, and production deployment via a wizard. The skill provides pre-built tools for interacting with Slack (reading messages, fetching threads, joining channels) and integrates with the Workflow SDK for durable conversations and human-in-the-loop approvals, and AI Gateway for model selection.

2/28/2026

This post details how Avalara used Vercel's v0 tool to rapidly prototype and build AI-native customer workflows, including a patent-pending Chrome extension. It highlights how v0 translates plain language descriptions into working prototypes, accelerating the design and validation cycles and enabling the creation of new products within months. The post emphasizes the shift in Avalara's building process from traditional methods to iterative prototyping facilitated by v0.

2/24/2026

This post introduces a security architecture for agentic systems by defining four actors (agent, agent secrets, generated code execution, filesystem) and advocating for distinct security boundaries between them. It proposes separating agent compute from generated code execution using microVMs and credential brokering proxies to mitigate risks from prompt injection and credential theft. Architectures discussed include zero boundaries, secret injection without sandboxing, and separating agent compute from sandbox compute.

2/9/2026

This post details the implementation of an AI Engine Optimization (AEO) system to track how coding agents discover, interpret, and reference Vercel's web content. It introduces Vercel Sandbox for isolated execution of agent CLIs, including installation, credential injection, and transcript capture. It also highlights the use of Vercel AI Gateway for routing LLM calls and centralizing management. A four-stage normalization layer is described to handle diverse agent transcript formats, including parsing tool names and flattening message structures into a unified event type.

2/3/2026

This post details the evolution of v0 from a novelty "vibe coding" tool to a production-ready platform. Key technical contributions include: a sandbox-based runtime that imports GitHub repos to work on existing codebases, automatically pulling environment variables and configurations from Vercel; a Git panel that allows for creating branches, opening PRs, and deploying on merge, enabling non-engineers to ship production code through proper Git workflows; secure integrations with Snowflake and AWS databases for building custom reporting and data apps; and built-in enterprise-grade security features like deployment protection and access controls, leveraging Vercel's platform security.

1/28/2026

This post details how Stripe used v0 to build a customer-facing value calculator application in a single flight. It highlights the adoption challenge of their previous spreadsheet-based models and explains why v0 was chosen for its ability to synthesize intent and generate deployable code. The post quantifies the impact of the v0 app, showing a 288% increase in internal users and significant time savings. It also discusses the broader impact on other teams at Stripe, inspiring them to build prototypes and demos.

1/27/2026

This post introduces and validates a new method for providing framework-specific knowledge to AI coding agents: embedding a compressed documentation index directly into `AGENTS.md`. This approach achieved 100% pass rates in evaluations for Next.js 16 APIs, outperforming the 'skills' abstraction which maxed out at 79%. The post details the problem of outdated training data, the two approaches tested (skills vs. `AGENTS.md`), the development of a hardened eval suite, and the surprising results. It explains the benefits of passive context (no decision point, consistent availability, no ordering issues) and addresses context bloat concerns through aggressive compression (8KB). A codemod is provided to automate this setup for Next.js projects.

1/26/2026

Introduces 'Agent Skills' as a new mechanism for providing agents with context, instructions, and resources. Defines skills as a standardized, open format (agentskills.io) for packaging reusable expertise. Differentiates skills from MCP servers, tools, rules, and system prompts, highlighting their strength in packaging complete workflows for complex, multi-step processes. Provides examples of skills (Vercel React Best Practices, Supabase Postgres Best Practices, Copywriting, Remotion Best Practices) and explains the concept of skill packages. Details installation methods (CLI, local placement) and how agents utilize skills by loading them based on task context.

1/22/2026

This post evaluates the effectiveness of different abstractions (SQL, bash, filesystem) for AI agents querying structured data. It details an evaluation harness, compares the performance and accuracy of SQL, bash, and filesystem agents, identifies performance bottlenecks in bash tools (e.g., `stat()` calls), and proposes a hybrid approach combining SQL and bash for improved accuracy and self-verification. It also highlights the importance of robust evaluation methodologies and the iterative improvement of tools like `just-bash`.

1/7/2026

Introduced a multi-step agentic pipeline for the v0 coding agent, comprising a dynamic system prompt, LLM Suspense for real-time streaming manipulation, and autofixers. The dynamic system prompt injects up-to-date knowledge about SDKs and provides filesystem examples. LLM Suspense handles tasks like URL shortening and icon name resolution during streaming. Autofixers address more complex issues by analyzing ASTs and using fine-tuned models for repairs like wrapping hooks and completing dependencies.

12/27/2025

Introduces the v0 AI agent, which allows business users to build and deploy internal tools by describing their requirements. These tools are built on the Vercel platform, leveraging its secure-by-default environment, automatic access control, secure integration capabilities, and guardrails against accidental exposure. The post highlights examples of internal tools built with v0, such as sales forecasting, marketing lead list import, and custom event marketing pages, and emphasizes the simplified maintenance and reversibility of changes for these applications.

12/22/2025

This post details a significant architectural simplification for Vercel's text-to-SQL agent, 'd0'. The previous complex system with numerous specialized tools, heavy prompt engineering, and context management was replaced with a file system-based agent that leverages standard Unix utilities (`grep`, `cat`, `ls`) to interact with the Cube semantic layer. This change resulted in a 3.5x faster execution time, 37% fewer tokens used, and a 100% success rate across benchmarked queries, highlighting the benefits of 'addition by subtraction' and trusting the model's reasoning with well-structured data.

12/15/2025

This post details the integration of v0 (an AI agent for building UI) with Notion. This allows v0 to access and utilize data from Notion databases and documents to generate dashboards, prototypes, and workflows. The integration also supports write access, enabling v0 to update Notion pages with generated content or summaries. This enhances the agent's ability to ground its output in existing team context and contribute back to knowledge bases.

12/15/2025

Introduces a structured prompt framework for v0, comprising three key inputs: product surface, context of use, and constraints & taste. Provides examples and test results demonstrating how this framework leads to faster generation times, smarter UX decisions, and cleaner code. This directly impacts the agent's ability to generate more accurate and efficient UI components.

11/24/2025

This post details the technical implementation of the v0 iOS app, focusing on its chat experience built with React Native and Expo. It covers composable chat architecture using context providers and custom hooks, animations for message entry and streaming using React Native Reanimated, keyboard handling with React Native Keyboard Controller, and native UI element integration. It also discusses code sharing between web and native platforms.

11/21/2025

This post introduces the concept of 'self-driving infrastructure' where AI agents autonomously manage production operations, improve application code using real-world insights, and learn from production behavior. It builds upon framework-defined infrastructure and the Vercel AI Cloud, with Vercel Agent as the first step towards fully autonomous cloud operations. Agent investigates anomalies, runs observability queries, performs root-cause analyses, and conducts AI code reviews in isolated sandboxes before deployment. It aims to close the production-to-code feedback loop by generating pull requests based on production data to improve stability, security, and performance.

11/6/2025

This post introduces a methodology for identifying and building high-ROI AI agent projects, focusing on tasks with low cognitive load and high repetition. It provides detailed workflows for a lead processing agent (deep research, qualification using generateObject, email composition, human review via Slack, approval/send) and an anti-abuse agent (URL intake, analysis, recommendation, human review, resolution). It also lists open-sourced agent templates and a hands-on program for agent implementation.

10/31/2025

Introduces the 'Investigations' skill for Vercel Agent, which automates incident response by detecting anomalies, performing AI-powered root cause analysis (correlation, historical context, dependency mapping, change attribution), and providing actionable recommendations. This enhances the agent's ability to act as a first responder to production issues.

10/23/2025

Introduces Vercel Agent, an AI teammate for development workflows, with two core skills: Code Review and Investigations. Code Review integrates with GitHub, analyzes code diffs, dependencies, and framework patterns, and uses Vercel Sandbox for simulated builds to validate recommendations before generating pull request feedback. Investigations are triggered by application anomalies, analyzing runtime data to identify root causes of issues and suggest next steps. The post also positions Vercel Agent as the first product in an 'Agent-as-a-Service' category, emphasizing secure, production-grade agents that are integrated into the workflow without requiring self-hosting.

8/22/2025

This post introduces the concept of AI-native design systems and how they enable more effective AI prototyping with tools like v0. It highlights the importance of design systems providing context to AI models, focusing on `shadcn/ui` as an example of a design system built with AI in mind. Key aspects discussed include open components, composable APIs, token-based styling, and the use of registries for sharing and contextualizing design systems for both humans and machines. It also touches on how v0's design mode and the Model Context Protocol (MCP) leverage these AI-native design systems.

8/20/2025

Introduces a new convention, `<script type="text/llms.txt">`, for embedding LLM instructions directly within HTML responses. This convention leverages the browser's behavior of ignoring unknown script types to embed LLM-specific content without affecting regular rendering. The proposal is based on the `llms.txt` standard and aims to provide agents with in-band instructions on how to access protected resources, such as Vercel preview deployments, by referencing available MCP server functions or bypass token mechanisms.

8/20/2025

This post details how Code and Theory leverages v0, a prompt-driven tool, to accelerate their prototyping, requirements gathering, and project delivery processes. It highlights the shift from traditional documentation to direct prompt-to-code workflows, resulting in a 50-75% reduction in deployment horizons and increased delivery speed. The post emphasizes the accessibility of v0 for cross-functional teams and its integration into production workflows via GitHub versioning, QA testing, and Vercel deployment.

8/11/2025

This post announces the rebranding of v0.dev to v0.app and highlights its evolution into an agentic AI builder. It emphasizes the capability of v0.app to go from a single prompt to a deployed full-stack application, including UI, content, backend, and logic. Key advancements include the AI's ability to research, reason, debug, and plan, handle complexity across all aspects of an app, and maintain security. New functionalities demonstrated include automatic error checking, design inspiration generation, web searching with citations, step-by-step planning, file reading, web inspection, to-do management, and integration implementation.

8/8/2025

This post details how Zapier uses v0's Platform API to programmatically generate personalized front-end mockups for its integration partners. This accelerates partner onboarding and increases the volume of partner demos by enabling the creation of high-fidelity prototypes in a few hours instead of days. The team is exploring ways to trigger mockups based on inbound partner interest, with a long-term vision of creating mockups for all 8,000 Zapier integration partners.

8/4/2025

This post details how v0, Vercel's AI-powered UI generation tool, incorporates security guardrails to prevent common AI-generated code vulnerabilities, such as exposed secrets and misuse of public variables, by applying automated, context-aware security checks during generation and before deployment. It also highlights Vercel's platform-level security features like the Vercel Firewall and secure inference via Vercel AI Gateway, ensuring secure defaults from prompt to deployment.

9/3/2024

This post details the integration of an AI-powered agent into Vercel's customer support workflow, reducing human-handled tickets by 31% through a comprehensive knowledge base and iterative rollout strategy. The agent handles a wide range of customer queries, transitions to human support when necessary, and provides valuable feedback for documentation and product improvements. The implementation involved A/B testing with Edge Middleware and a focus on customer experience and brand integrity.