update
Apr 3, 2026
By Teun
MCP Dev Summit NYC draws 1,200 attendees and previews stateless transport
The first MCP Dev Summit under the Agentic AI Foundation drew about 1,200 developers to New York on April 2-3. Key announcements included stateless transport work via SEP-1442 to fix load balancer compatibility, MCP Apps adoption across Claude, ChatGPT, VS Code and Goose, and real-world case studies from Amazon, Uber, and Duolingo. Duolingo shared how they went from a few engineers to 250 weekly active MCP users internally.
The first MCP Dev Summit under the Agentic AI Foundation brought about 1,200 developers to New York on April 2 and 3, and the agenda was less about hype than about making the Model Context Protocol fit real production environments. The biggest technical item was work on stateless transport, discussed through SEP-1442, which is intended to improve compatibility with load balancers and other infrastructure that expects requests to be handled without relying on long-lived server sessions.
MCP, or Model Context Protocol, is a standard for connecting AI assistants and agents to external tools and data sources. Instead of building one-off integrations for every app, teams can expose capabilities through MCP servers and let compatible clients discover and use them in a consistent way.
⚡ New to this?
This is about MCP, short for Model Context Protocol, a standard that lets AI apps connect to external tools and data in a common way. A lot of AI systems today need custom wiring to talk to databases, apps, or internal services, so a shared protocol can make those connections easier to build and maintain.
The stateless transport work matters because it affects how MCP services run behind normal infrastructure like load balancers, which spread traffic across multiple servers. The summit also showed that MCP is moving beyond experiments, with major clients and companies starting to use it in real products and internal workflows.
🦞 OpenClaw angle
If you run MCP servers including Basic Memory or custom integrations, watch for the stateless transport change in SEP-1442. It will affect how remote MCP servers handle sessions and may require updates to proxy setups. The Duolingo case study of scaling from a handful of users to 250 weekly active MCP users is worth reading for adoption patterns.
That standard has moved quickly from early adoption to broader ecosystem work. At the summit, organizers and speakers highlighted MCP Apps support across several major clients, including Claude, ChatGPT, VS Code, and Goose, which points to a growing expectation that MCP will become a common layer for agent tooling rather than a niche protocol for experimenters.
The transport discussion matters because many real deployments sit behind proxies, load balancers, and other network controls. Traditional session-based approaches can be awkward in those environments, especially when traffic is distributed across multiple instances, so a stateless model is meant to make remote MCP servers easier to run at scale.
The event also framed MCP as a practical enterprise tool rather than a demo-only standard. Presentations and case studies came from companies including Amazon, Uber, and Duolingo, showing how teams are applying the protocol inside products and internal workflows.
Duolingo’s example stood out because it showed how quickly internal usage can grow once a tool is useful to engineers. According to the summit summary, the company described moving from only a few engineers using MCP to 250 weekly active internal users, which suggests that developer-facing infrastructure can spread fast when it lowers the cost of connecting systems.
That kind of internal growth is one reason MCP has attracted attention from teams building AI automation systems. If a protocol can standardize how models reach tools, it can reduce repeated glue code, make integrations easier to reason about, and create a more portable approach across agents and clients.
The summit itself also marked an organizational milestone. By hosting the first MCP Dev Summit under the Agentic AI Foundation, the ecosystem is signaling that MCP is moving from a loosely coordinated community effort toward a more formal development track, with public discussion around transport, client support, and operational patterns.
