April 2025 Summaries

The term "ephemeral" in the database world refers to something temporary, short-lived, and not meant to persist. This concept is particularly useful for branches that need a fresh, isolated copy of a database for previews, CI test runs, or local development. Neon Local is a proxy service that creates a local interface to a Neon cloud database, automatically creating a new branch when the container starts and deleting it when it stops. It adopts a hybrid approach by using a combination of cloud-based, isolated environments and Docker containers to provide a truly ephemeral experience. Neon Local can be used with various Postgres clients and configuration options, such as parent branch IDs and delete branch settings, to suit different needs. The service is designed to simplify working with branches in Docker environments, making it easy to develop, test, or preview features in isolated, clean environments that feel local but behave like production.

Neon Local is a proxy service designed to facilitate the creation of ephemeral environments using Docker, which allows for temporary, non-persistent database branches that automatically create and delete with each container's lifecycle. This service enables developers to maintain CI-friendly branching by handling the automatic creation and removal of database branches during test runs, supports local development by streamlining branch management without manual configuration, and ensures consistent connectivity by routing applications through a local Postgres endpoint. Although Neon Local is not a local Postgres instance, it offers advantages over traditional local setups by preventing environment drift and simplifying the process of spinning up production-like environments. It supports various configurations, including setting a parent branch ID or deciding whether to delete branches upon container shutdown, and future features may include the ability to assign meaningful names to branches and an offline mode for local database dumps and restores.

If you're building an app using AI code generation, it's recommended to use Neon for your database, as its built-in safety features ensure security and smooth operation. Neon offers a free tier with limited databases and branches, and has integrations with popular services like v0, Windsurf, and create.xyz. The platform provides protection against malicious attacks, recovery options if something goes wrong, and helps prevent SQL injection by safely parameterizing variables. It also includes features like Point-in-Time Restore (PITR), Snapshots, protected branches, IP allowlisting, secure connections, and password reset. By using Neon, developers can minimize risks, avoid costly mistakes, and focus on building their app without worrying about potential threats.

Neon offers a robust solution for database management, emphasizing security and resilience particularly in applications developed using AI code generators. Key features include built-in safety mechanisms such as Point-in-Time Restore (PITR) and Snapshots for data backup and recovery, SQL injection prevention through parameterized queries, and protected branches to safeguard against accidental deletions. The platform ensures secure connections with SSL/TLS encryption and allows IP allowlisting to restrict access to trusted sources. Additionally, Neon simplifies password management and maintains secure database connections, enabling developers to focus on coding confidently without worrying about potential vulnerabilities or data breaches.

The article explores the challenges of managing AI agent data access securely, particularly in multi-user environments where different privilege levels exist. It discusses various implementation approaches to ensure users only see data they are authorized to access, including base scenarios, role-based access control, middleware enforcement, row-level security, proxy database access, API gateway policies, token-based access control, and data masking. The article emphasizes the importance of proper authentication, API security, database-level filtering, and AI output moderation to build secure AI-driven applications that comply with data governance policies and protect sensitive information from unauthorized access.

Designers often face frustrating UX papercuts that erode product quality, but development teams are focused on bigger priorities. Vibe coding with tools like Devin empowers designers to break this cycle by leveraging AI codegen, enabling them to "vibe code" away the backlog and ship real improvements without waiting in the queue. Devin is an AI-powered code generation tool that understands system architecture and repo structure, allowing designers to prompt it with scoped changes and get back a ready-to-review pull request. By using Devin, designers can quickly scaffold solutions and handle foundational work, giving them more capability and freeing up engineering bandwidth. This shift empowers designers to raise the quality bar, act on design intent, and deliver a smoother experience for users.

As AI agents become increasingly integrated into applications ranging from small demos to enterprise-level solutions, managing secure data access poses significant challenges. These AI-driven applications require robust access control mechanisms to ensure that users with varying privilege levels only access data they are authorized to see. The discussion highlights scenarios where access is controlled at multiple levels, such as database queries, API calls, and AI-generated outputs. For instance, User A has unrestricted access to resources, while User B's access is limited by role-based constraints, ensuring they cannot retrieve sensitive data without proper authorization. Strategies like row-level security, API gateway policies, and data masking are employed to enforce these controls. The implementation of role-aware AI agents dynamically adjusts based on user permissions, ensuring compliance with data governance policies and protection of sensitive information. The article underscores the importance of combining authentication, API security, and database filtering with AI output moderation to build secure AI-driven applications.

AI tools like Devin are transforming the workflow of designers at Neon by allowing them to directly address UX papercuts, which are minor yet persistent interface issues, without being hindered by development backlogs. Devin facilitates "vibe coding," enabling designers to generate production-ready code and submit pull requests, thus expediting the improvement process and enhancing product quality. Although Devin excels at smaller tasks, it requires more oversight for extensive changes, akin to a junior developer. This capability empowers designers to work alongside engineers more effectively, contributing to a more agile and responsive design system. Additionally, by integrating AI-driven tools like V0 and Lovable, Neon is streamlining its research process, allowing for quicker iterations and more informed decisions, ultimately fostering a collaborative culture where designers are active contributors to product development. Neon’s serverless Postgres database, compatible with AI IDEs and platforms like Replit Agent, further supports this agile environment.

Most developers approach AI coding by giving the AI a one-line prompt, expecting it to produce high-quality results without much guidance. However, this approach often leads to hallucinations, slop, and edge cases missed. A better approach is to treat the AI like a senior developer, providing context, reviewing work, iterating, and asking for deeper thinking. This involves using tactical prompting, understanding that LLMs shine when treated as a senior teammate, and giving clear instructions with well-defined requirements. Additionally, it's essential to allow the AI to ask questions, iterate with the AI, and get maintainable, idiomatic code. By taking this strategic approach, developers can achieve far better results and move from producing programming to producing products.

To effectively harness AI in coding, it is recommended to treat AI like a senior developer rather than a junior one. This approach involves providing AI with comprehensive, structured input similar to a product requirements document (PRD), allowing for better tactical prompting and enabling the AI to act as a collaborative partner in development. By encouraging AI to ask clarifying questions and iteratively work through tasks, developers can leverage AI's strengths in exploring architectural decisions, testing strategies, and identifying potential pitfalls. This method reduces the time spent on fixing AI-generated mistakes and enhances the quality and reliability of outputs. Additionally, prompting AI to explain its decisions and identify uncertainties fosters transparency and allows developers to focus on critical areas needing verification. Adopting this mindset not only improves the quality of AI-generated code but also aligns it with best practices and team standards, resulting in more maintainable and idiomatic solutions.

Snapshots in Neon provide an easy way to capture a backup of your database at a specific point in time, offering a lighter alternative to traditional backup solutions. They are designed to preserve the exact data and schema at the moment they were created, making them ideal for testing, debugging, or rolling back to a known good state. Snapshots can be conveniently restored into a new branch without altering the original branch, complementing traditional backup strategies but offering an easier, more flexible recovery option when needed. They are available to all users in Neon's Early Access Program and provide instant backups with minimal performance impact. The program also includes support for creating custom retention periods and configuring snapshots on schedule, as well as programmatic control via API and CLI.

The State of AI 2025 survey polled thousands of developers to understand how AI tools fit into real workflows. The survey found that some AI coding tools are gaining mainstream adoption, but there are still barriers to their use. Cursor is winning among developers for fast prototyping and refining output with manual reviews, while v0 is the most widely used codegen tool with high positive sentiment. Model providers like Open AI's ChatGPT and Copilot remain dominant in the market, capturing 91.2% of total awareness for ChatGPT. The survey highlights both the benefits and challenges of using AI tools in development workflows, including issues with code quality and editor lock-in.

Neon has introduced Snapshots, a feature that allows users to create read-only, point-in-time copies of their databases, aimed at enhancing disaster recovery, testing, and rollbacks. These snapshots offer a lightweight and instant alternative to traditional backup methods, enabling users to capture the exact state of their database without significant performance impacts. Available through Neon’s Early Access Program, snapshots can be restored into new branches for testing or debugging without altering the original branch, providing a flexible recovery option that complements traditional backup strategies. Unlike Point-In-Time Restore (PITR), which reverts entire branches to a previous state, snapshots are ideal for creating rollback points and facilitating long-term data retention for compliance purposes. Neon's future plans include adding features such as custom retention periods, scheduled snapshots, and programmatic control via API and CLI, enhancing the versatility and reliability of their database management solutions.

The first-ever "State of AI 2025" survey, organized by Devographics, gathered insights from over 4,000 developers worldwide to understand the integration of AI tools into coding workflows, highlighting the evolving role of AI in software development. The survey found that AI coding tools like Cursor and v0 are gaining traction, with Cursor being notably favored for its deep integration and autocomplete features in IDEs, while v0 leads in code generation tools, praised for its positive impact on productivity. Despite the increasing adoption, developers remain cautious about AI-generated code quality, emphasizing the need for manual reviews and expert oversight. OpenAI's ChatGPT maintains a dominant presence as a model provider, while GitHub's Copilot remains a leading coding assistant, though Cursor is emerging as a potential alternative. Overall, AI is becoming integral to development processes, although challenges such as code quality and editor preferences persist, indicating that human expertise remains essential in AI-assisted coding.

Structured prompting is an emerging critical skill for modern developers to master, enabling them to effectively use AI as a coding partner. By understanding how to craft effective prompts and leveraging various techniques such as few-shot learning, chain-of-thought reasoning, self-consistency, skeleton prompting, output schemas & format directives, configuration parameters, prompt anatomy, linting, and LLM-powered self-linting, developers can harness the power of AI to generate high-quality code while maintaining control over structure, determinism, creativity, and consistency. This approach mirrors traditional software development practices, such as testing and version control, allowing developers to confidently ship AI-powered features and build robust, reliable systems.

Prompt engineering is emerging as a critical skill for developers as AI becomes increasingly integrated into software development, functioning as an ideal pair-programming partner. Understanding how to craft effective prompts is vital for leveraging AI to generate high-quality code, akin to software engineering practices like DRY and SOLID. This involves treating prompts as modular, testable components that can be refined and integrated into larger systems. Structured prompting, including techniques like few-shot learning, chain-of-thought reasoning, and self-consistency checks, enhances the quality of AI-generated outputs by providing clear examples and logical progression. Developers are encouraged to use skeleton prompts and define output schemas to ensure adherence to coding standards and system compatibility. Furthermore, prompt linting and configuration parameters, such as temperature settings, allow developers to validate and tune prompts for optimal performance. By adopting these methodologies, developers can build, test, and maintain prompts with the same rigor as traditional code, resulting in reliable AI interactions and reducing the unpredictability often associated with AI-generated content.

Neon offers the most powerful Postgres API on the market, enabling efficient and scalable database provisioning and management. Its serverless architecture and experience with embedded Postgres workflows make it a natural fit for platforms like Retool and AI agents like Replit Agent. The Neon API provides fine-grained control over provisioning, scaling, and usage, allowing teams to manage large fleets of Postgres databases programmatically while staying cost-efficient. It also offers features such as consumption quotas, compute settings, and real-time tracking of usage across thousands of projects. Additionally, Neon has dedicated interfaces for AI agents, including the Model Context Protocol (MCP) server, which enables agents to interact with APIs using structured natural-language commands. The API is useful for managing Postgres fleets at scale, particularly in platforms and AI agents that deploy thousands of isolated databases.

Neon offers a robust Postgres API designed for platforms and AI agents that require scalable and efficient database provisioning and management. Its serverless architecture supports instant provisioning and scale-to-zero, making it ideal for applications like Replit and Create.xyz, which need to deploy thousands of databases daily. The Neon API provides fine-grained control over database fleets, allowing developers to set consumption quotas, manage autoscaling, and track usage in real time. Additionally, Neon introduces the Model Context Protocol (MCP) server, enabling AI agents to interact with APIs through structured natural-language commands, enhancing autonomous database management. The toolkit includes a TypeScript SDK and a Serverless Driver for seamless integration into edge environments, offering developers an efficient means to manage extensive Postgres projects programmatically.

This modern startup stack on Azure is designed to help early-stage startups build and deploy a scalable, secure, and efficient application. The stack consists of a simple core architecture that includes Azure Entra ID for user authentication, Azure DNS + Application Gateway for routing traffic, Azure Virtual Network for isolating internal resources, Static Web App + App Service Plan for hosting the frontend and backend API, Neon Serverless PostgreSQL as a modern relational database, Azure Blob Storage for storing raw customer files, Azure Key Vault for managing secrets and credentials, GitHub Actions + Container Registry for automating build, test, deploy processes, and Azure Monitor + Log Analytics for logging, diagnostics, and alerts. The architecture is designed to scale horizontally with the addition of new customers, data, and usage, while maintaining security and control throughout.

The text outlines a modern startup stack architecture on Azure, emphasizing its suitability for fast-paced, early-stage startups aiming to test ideas and keep costs low. It presents a case study of a startup building a Retrieval-Augmented Generation (RAG) application that uses Azure services to manage multiple customers' private data and AI experiences. The stack includes Azure services like Azure Entra ID for user authentication, Azure DNS for traffic management, Azure App Service for backend and frontend hosting, and Neon Serverless PostgreSQL for scalable data storage. It highlights the benefits of using Azure's managed services to ensure scalability, security, and cost-efficiency while allowing startups to focus more on product development. The architecture supports features like multi-staging environments, automatic deployments with GitHub Actions, and global reach with Azure Front Door, all designed to facilitate growth and maintain security as the startup expands its customer base.

When a Postgres failure occurs, the time it takes to recover can have significant consequences for businesses, including lost revenue, customer churn, and even breaches of Service Level Agreements. A recent survey found that nearly 30% of teams experienced more than three hours of downtime, while only 21% felt confident in their ability to quickly restore after a failure. The survey also revealed that 52% reported negative customer feedback and a spike in support cases due to the incident, highlighting the importance of reliable database restores. However, many companies are not adequately prepared for such events, with 68% having faster point-in-time recovery solutions like Neon's on their wishlist. Neon offers instant PITR (Point-In-Time Recovery) based on branching, eliminating the need for replaying WAL or waiting for a full database restore, and providing instant restores even at large scales.

The Neon MCP server is a remote version of the local MCP server that allows large language models (LLMs) and AI agents to interact with external systems in a structured way. The remote server uses Server-Sent Events (SSE) for communication and integrates OAuth 2.1 for authentication. The implementation involves using an Express.js web server to handle HTTP connections for SSE and OAuth, and implementing the OAuth flow with Neon as the upstream provider. The server also supports both stdio and SSE operations, and refactors the code to introduce a pattern where each handler function receives a neonClient parameter. The OAuth 2.1 authorization flow allows users to connect and authorize the MCP server to act on their behalf without manually sharing API keys. The remote MCP server is live and can be installed on Cursor or Windsurf, but it's still at an early stage with known issues such as dependency on Node.js and refresh tokens.

Neon has developed a remote Model Context Protocol (MCP) server that allows AI agents to interact with the Neon Postgres platform using natural language commands. The server uses Server-Sent Events (SSE) for communication and integrates OAuth 2.1 for secure authentication, addressing previous issues with manual installation and API key management. The remote server architecture leverages SSE for server-to-client messages and HTTP POST for client-to-server communication, enabling AI agents to execute commands like creating projects or managing databases. The server operates as both an OAuth client and server, allowing users to authorize the MCP server without sharing API keys directly. Deployment was facilitated using a Docker image on Fly, with the potential for exploring other platforms in the future. Although the remote server simplifies installation, challenges remain, such as dependency on Node.js and handling refresh tokens, which are being addressed to enhance user experience and broaden adoption.

Slow point-in-time recovery (PITR) in PostgreSQL services, such as AWS RDS, presents a significant challenge for companies, leading to extended downtimes, customer dissatisfaction, and potential revenue losses. A survey of 50 developers managing large Postgres databases revealed that 59% experienced critical failures in the past year, with nearly 30% facing downtimes exceeding three hours. These incidents often result in negative customer feedback, SLA breaches, and disrupted engineering workflows. To address these issues, Neon offers a novel approach to Postgres recovery with instant PITR, allowing users to quickly revert to any past state without the lengthy wait associated with traditional methods. Neon's solution eliminates the need for lagging replicas and costly standby instances, providing a more efficient and scalable recovery process, which is particularly beneficial for businesses where uptime is crucial.

The Keycloak plugin extends the built-in email change functionality to notify external systems when a user changes their email address, ensuring data consistency and preventing authentication issues. The plugin consists of two main components: a REST endpoint for initiating the email change request and a token handler for processing the verification link, verifying the email update, and updating external systems accordingly.

Adi Griever, a Back End Golang Developer, details the creation of a custom Keycloak plugin designed to handle email change events and synchronize external systems efficiently. The plugin addresses the challenge of keeping systems in sync when users update their email addresses, a feature not natively supported by Keycloak. Implemented as a Keycloak Service Provider Interface (SPI), the plugin consists of a REST endpoint and a token handler, which work together to initiate email change requests, verify the new email, and ensure data consistency by updating external systems. The process involves generating and verifying an email verification token, unlinking existing social provider links to avoid authentication issues, and notifying external services of the change. This solution effectively manages email updates while maintaining security and reducing manual intervention, showcasing Keycloak's extensibility.

Vibe coding`, a relatively new concept, has emerged in the developer community, referring to AI-assisted workflows where developers let models generate full chunks of their app and roll with it. However, as projects grow, this approach can hit its limits, leading to bugs, untraceable changes, and introduced dependencies. To address this, Cursor Tab has been developed, an "autocomplete on steroids" feature that enables "tab coding," a human-AI collaboration approach where developers guide the AI through intentional implementation steps. Tab coding uses predictive coding and zero-entropy edits to reduce repetition, maximize predictive intelligence, and enhance user experience by providing intuitive UI elements. The technology combines advanced AI and optimization techniques, including hierarchical model architecture, low latency optimizations, and intelligent processing methods, such as speculative edits and caching-aware prompting. By empowering developers beyond vibe coding, Tab aims to maintain agency while eliminating implementation friction, providing knowledge rather than just code.

At Neon, we like to be on top of things, especially when it comes to AI. We first announced our MCP server back on December 3rd, 2024, and have been iterating on its capabilities since then. A remote MCP Server greatly simplifies setting up the server through any client without having to create API keys in our service. Our hosted version is now available for users to try, but it's launched under a preview state due to the new OAuth specification still being very new. To use the remote MCP Server, simply register a new MCP Server in your client's settings and follow the instructions provided, which can be found on our website. The server communicates with clients via SSE and can be deployed on many different cloud providers. We're committed to refining and expanding our offering as the MCP specification evolves, making AI workflows more accessible, scalable, and future-proof.

The author created an interactive timeline showcasing major Postgres features and releases, highlighting the power of dedicated development over 30 years. They utilized Replit to build a website scraper that efficiently gathered release notes from Postgres' official website, processing over half a million words. The data was then transformed using OpenAI's ETL script before being visualized with Vercel v0, an AI agent capable of creating attractive and interactive frontends for Next.js applications. The final product is a dynamic website celebrating Postgres innovation, made possible by combining Replit, Vercel v0, and Neon Database.

Exploring the extensive history of PostgreSQL, a project was undertaken to create an interactive timeline showcasing major features and releases over the past 30 years, utilizing Replit and Vercel v0. The process began with Replit's AI agents and web scraping tools to gather data from PostgreSQL's official site, capturing information from 523 releases across 33 major versions. This data was then processed using an ETL script and the OpenAI API to distill key features, all stored in the Neon Postgres database. Vercel v0, with its expertise in modern web technologies and Next.js, was employed to create an engaging and functional frontend, including a Gantt chart illustrating the lifecycle of PostgreSQL versions. The entire project was executed through natural language commands, resulting in an interactive website that vividly presents PostgreSQL's developmental history, hosted free on Vercel.

Cursor Tab, a feature within the Cursor coding environment, offers a more structured and developer-centric alternative to the rapidly popularized "vibe coding," which relies heavily on AI to generate code. Cursor Tab facilitates "tab coding," a workflow where developers and AI collaborate to streamline coding tasks by using predictive intelligence to anticipate developers' next steps, thereby minimizing repetitive tasks and enhancing user experience. This method focuses on zero-entropy edits, allowing developers to express their intent with the AI handling predictable coding actions, resulting in a more efficient and controlled coding process. Unlike vibe coding, which can lead to untraceable changes and bugs, tab coding provides a more intentional and predictable acceleration of workflows, enabling developers to maintain agency and focus on higher-level decisions while the AI assists with implementation details. Cursor Tab employs sophisticated AI models and optimizations to ensure a responsive, intuitive experience, supporting developers in maintaining consistency across their codebase and adapting to individual coding patterns.

Neon has announced the launch of its remote Model Context Protocol (MCP) server, which facilitates AI workflows by simplifying server setup through clients like Cursor or Windsurf without requiring API keys. This development follows the company's December 2024 introduction of the MCP server, with ongoing iterations to enhance its capabilities. The remote server, accessible via https://mcp.neon.tech, automatically updates with new features, eliminating the need for local upgrades. Currently in a preview state due to the nascent MCP specification for OAuth, Neon anticipates potential setup changes and initial instability. The server is open-source, using the geelen/mcp-remote Node package, and communicates with clients through server-sent events, deployable across various cloud providers. As the MCP specification progresses, Neon aims to enhance the server's reliability, making AI workflows more accessible and scalable, and invites user feedback via Discord.

The noisy neighbor problem in cloud architecture and databases refers to the challenge of managing competing workloads from multiple tenants sharing the same resources, such as CPU, memory, and I/O. This can lead to performance degradation for other tenants due to disproportionate resource usage by one or more tenants. The problem arises when a single tenant's workload consumes excessive resources, degrading performance for all other tenants sharing the same infrastructure. To address this issue, database architects must design systems that can accommodate varying workload patterns without allowing any single tenant to disrupt service quality for others. Isolation strategies, such as resource governance, schema separation, and complete isolation, can help manage noisy neighbors in shared models like AWS RDS. However, these approaches often come with drawbacks, including cost inefficiency, operational complexity, and maintenance burdens. Neon, a serverless, disaggregated database architecture, provides an elegant solution to the noisy neighbor problem by structurally eliminating resource contention between tenants while maintaining flexibility and alignment with the SaaS business model.

In cloud-based multitenant architectures, the "noisy neighbor" problem arises when one tenant's excessive use of shared resources, such as CPU, memory, or disk I/O, degrades performance for other tenants. Traditional solutions like AWS RDS struggle with this issue due to resource contention, prompting strategies like query limits, connection pooling, and workload segregation. However, these approaches often lead to operational complexity and inefficiency. Neon offers a more efficient solution by providing database-per-tenant isolation, decoupling compute and storage, and implementing autoscaling to ensure resource usage aligns with demand. This architecture minimizes contention, allowing tenants to scale resources independently, thus maintaining performance without the drawbacks of over-provisioning or operational burden. By aligning costs with actual usage, Neon effectively addresses the noisy neighbor problem, making it a promising solution for SaaS providers seeking to optimize their multitenant applications.

Neon has partnered with GitHub Secret Scanning to enhance the security of its users, automatically detecting exposed Neon database credentials and API keys in public repositories and npm packages. This integration provides an additional layer of protection against unauthorized access, ensuring that exposed credentials are identified quickly and mitigated before they can be abused. To prevent accidental leaks, Neon recommends following security best practices such as using environment variables instead of hardcoding credentials and storing sensitive information in secret management tools. The partnership reinforces Neon's commitment to security and provides users with a proactive approach to securing their credentials.

Neon has joined GitHub's Secret Scanning Partner Program to bolster database security by automatically detecting and addressing exposed database credentials and API keys in public repositories. This collaboration ensures that when GitHub identifies Neon credentials, an automated system alerts both Neon and its users to mitigate potential risks. The partnership underscores Neon's commitment to security by offering an additional layer of protection against unauthorized access. Users are encouraged to adopt best practices, such as using environment variables, secret management tools, and regularly rotating credentials, to further secure their data. Neon continues to prioritize security enhancements and invites users to reach out for support regarding this integration.

Velocity, a fast-moving startup, switched from AWS RDS to Neon for their database solution due to easier scalability across clouds, allowing them to scale effortlessly without managing separate configurations per cloud. This switch also provided faster development workflows through Neon's branching model, which enabled developers to instantly create branches and test schema changes safely. Additionally, Velocity gained reliable support when it mattered most, with immediate access to real support engineers. They are now experimenting with pgvector for embeddings and leveraging observability features to diagnose issues quickly.

Velocity transitioned from AWS RDS to Neon to enhance scalability, flexibility, and simplicity in their AI-powered production engineering platform, which aims to improve incident response time by replacing manual workflows with intelligent automation. Neon offers a serverless, multi-cloud architecture that allows for effortless scaling without dedicated infrastructure management, supporting both AWS and Azure. This switch has led to faster development workflows through Neon's branching model that enables lightweight database clones for safe testing and experimentation. Additionally, Neon provides robust support, real-time database insights, and the ability to handle vector search with pgvector, consolidating Velocity's data stack and optimizing performance without the complexity of multiple specialized databases. This strategic move aligns with Velocity's mission to build a cloud-agnostic AI Production Engineer platform, ensuring reliable support and efficient operations.