Speakeasy

Great companies move quickly. So shoutout to the companies that are hopping on the MCP wave and rolled out servers with us this week! Check out their servers in the comments 👇 Dub.co‎‎, Polar‎, Hathora‎, Novu‎, Acuvity‎, Stream Store, Ava Labs‎‎‎

I'm excited to announce the Event Destinations Initiative. A community effort and guidelines to promote better event interoperability between producers and developers. While webhooks benefit from flexibility, they also present serious issues. Event destinations maintain webhook support and enable improved security, developer experience, and performance while reducing operational costs. Stripe coined the term "Event Destinations" at Sessions last year, and this Initiative destination builds on the work of industry leaders such as Shopify and Twilio that have supported event delivery to alternative event bus & gateways for years. I'll be sharing more insights on this topic in the coming weeks. If you're interested in contributing, providing feedback, or implementing event destination patterns in your service, please DM me. Special thanks to Patrick Malatack Vlad Pick Sagar Batchu 🚀 David Boyne Lauren Long Alex Plugaru Phil Leggetter Paul Asjes and many others whose support made this possible. Link to the full documentation and resources in comments! ⬇️

The power of APIs... One team solves a problem for an entire industry. Thanks to Gusto‎ nobody needs to worry again about building payroll systems. With Gusto embedded you just write a few lines of code, and payroll works as expected. We're excited to be working with the Gusto team to make the integration experience even easier! More to come!

The question every engineering leader is responsible for answering in 2025: "How are we preparing our APIs for agentic workflows?" 👇 ‎ ‎For the last 20 years, it's been a given that API integrations would be built by developers, that the access patterns would be predictable, that well written API docs would be enough to enable users. In 2025, that's suddenly changed. Now integrations are being built by AI, the patterns are dynamic and ever changing and most importantly the means of discovering APIs and the mechanism for integrating are shifting. The space is changing every day, but right now the single most important artifact to make your API AI-ready is having a model context protocol (MCP) server. MCP servers are quickly becoming the defacto standard for giving AI agents access to API platforms. Without one, you risk losing out on a wave of API usage coming from AI tools like Claude, Cursor, Windsurf and many more. Long term as more application become agentic, having an MCP server will be business critical. That's why the Speakeasy TypeScript generator now bundles an MCP server into every TypeScript SDK — free of charge. Existing customers, just merge a PR New customers, just download the CLI and run quickstart. That's all you need to do to make your API platform readily accessible to LLMs and other AI platforms. This is Day 0, we're working on support for other languages as well as more agentic frameworks. When we started Speakeasy‎, our mission was to make every API easy to create and consume. We wanted to make API integration a breeze so that developers could spend more time building and less time wrangling integration code. That mission hasn't changed, but is has been extended to include the newest consumers of APIs. Man or machine, nobody should need to struggle with API integration!

🤖 How do you make your API AI-ready? With the latest acronym hot off the press — MCP: Model Context Protocol. It's about to permanently change how we build software 🚀 ‎ To truly unleash the power of AI, we need to enable AI to access real-time data and take actions on 3rd party systems — just like a human can. That means connecting to APIs. Different standards are emerging to enable this — including Model Context Protocol (MCP) from Anthropic. With tools like MCP, we can enable AI agents to move beyond static training data, by allowing them to make calls to APIs as part of their task execution. So how do you build an MCP server for your API? That’s where we come in. Speakeasy provides every API developer with a modern toolchain for REST API development. We generate idiomatic, rich docs, SDKs and tests for every REST API. Starting today, every Speakeasy-generated TypeScript SDK will come bundled with a Model Context Protocol (MCP) server — all built from your OpenAPI spec, no extra work required  What’s next? Check out the release post in the comments to learn how to generate an MCP server with Speakeasy! We have much more coming up in this space, so follow us to learn more about our plans

Our very own Tristan Cartledge is going to be speaking at the Tropic Tech meetup. If you're in Cairns, skip the Barrier Reef and go to see this talk instead!

The ongoing debate over HTTP GET with request bodies highlights a critical gap in REST APIs: how do we handle complex queries while staying true to REST principles? Currently, we see three distinct approaches in the wild: 1. GET with Query Parameters: The classic RESTful approach. Simple and cacheable, but becomes unwieldy with complex queries and runs into URL length limitations. 2. GET with Request Body: A controversial pattern used by some enterprise systems. While technically possible, it's discouraged by HTTP specifications, unsupported by fetch() API, and breaks caching semantics. Many API tools like OpenAPI generators explicitly reject this pattern. 3. POST for Complex Queries: A common workaround, but sacrifices the semantic benefits of GET – namely idempotency and cacheability. Some implementations use custom headers (X-HTTP-Method-Override) to indicate the intended semantics. This fragmentation in the ecosystem creates challenges for API providers, tooling vendors, and developers alike. That's why we're excited about the emergence of the HTTP QUERY method specification (currently in draft). QUERY is purpose-built for complex queries while maintaining the safety and idempotency of GET, but with explicit support for request bodies. What makes QUERY particularly promising is how it clarifies the semantic distinction between resource retrieval (GET) and search operations (QUERY). The specification also opens the door for better caching strategies for complex queries – something that's been a pain point in API design for years. We believe standardizing QUERY would be a significant step forward for the REST ecosystem. It would provide a clean solution for complex query operations while maintaining REST principles and enabling better tooling support. We're closely watching the specification's progress and hope to see wide adoption from frameworks, API gateways, and client libraries in the coming months. #APIDesign #REST #WebStandards #DeveloperTools

The LLM companies are making AI ubiquitous, Fireworks AI‎ is making it usable. Their focus on low latency, cost control, and high quality is helping companies transition from AI prototype to production use case. We couldn't be happier to be working with their team to create an API developer experience that's best in class!

"We're in day zero of the AI-API shift." Our CEO Sagar Batchu‎ joined the Stack Overflow Podcast to discuss how AI agents are changing API development: - The shift from developer-oriented to machine-to-machine communication - Why traditional API patterns might not work for AI - The future of API authentication - How companies can prepare for AI-first APIs Listen to the conversation: https://lnkd.in/gui-7SWH #APIDevelopment #AI #TechTalks

I'll be speaking at this month's PyBelfast Meetup at Slice HQ in Belfast on Wednesday. I'll be discussing how we are building a truly exceptional Developer Experience for our clients. Starting with a focus on consistent API design I'll show how that has led us to develop our first open-source libraries in Python (using Speakeasy) that significantly reduce the time taken for our clients to go live. https://lnkd.in/erYVaVsp #DeveloperExperience #OpenAPI #SDK