Kong launches Volcano, a cloud platform for building and running AI agents Kong launched Volcano, a cloud platform for building, hosting and running AI agents and web apps, at its API + AI Summit on September 30, 2026. Volcano bundles durable workflows with checkpointing and leader election, branchable PostgreSQL databases with vector support provisioned in seconds, and edge functions, and integrates with Claude Code, OpenAI Codex and Cursor; a free Hobby tier is available, with sandboxed compute arriving shortly after launch and general availability targeted for November 19, 2026. Kong, which raised a $175 million Series E in late 2024 at a $2 billion valuation and now calls itself "The AI Connectivity Company," is moving from API traffic management into hosting the runtime for production AI workloads. Kong launches Volcano, a cloud platform for building and running AI agents Photo: Tima Miroshnichenko / Pexels The API company best known for managing traffic now wants to host the AI agents generating it Kong has launched Volcano, a cloud platform built to help developers create, host and run AI agents and modern web apps from a single place. The company unveiled it on September 30, 2026, at its API + AI Summit. Kong now calls itself “The AI Connectivity Company.” What Volcano actually does Kong is pitching Volcano as a “batteries included” platform. Developers get workflows, storage and supporting services bundled into one cloud offering, so they spend less time assembling infrastructure. The first major piece is durable workflows . These support leader elections and checkpointing, two features that sound dry but matter a great deal for agents. Checkpointing lets a long-running process save its progress. If a step fails partway through, the work can resume from the last saved point. Leader election tackles a different problem. When several copies of a service are running, it picks one to be in charge, so they don’t trip over each other doing the same job twice. The second piece is branchable PostgreSQL databases . Kong says these can be provisioned in seconds and include vector support. AI, tech, and the markets they move—in one daily briefing. Daily. Free. Join 34,000+ readers across crypto, finance, and policy. Branching works much like it does in code. A developer can spin off a copy of a database, experiment with it, and leave the original untouched. Vector support matters because many AI applications store and search data as numerical representations of meaning rather than plain text. The third piece is edge functions . These are small chunks of code that run close to users, which helps keep response times quick and lets the system scale without much manual effort. Built for the AI coding crowd Volcano integrates with three popular agentic coding tools: Claude Code, OpenAI https://cryptobriefing.com/markets/openai/ Codex and Cursor. According to Kong CTO Marco Palladino, the integrations are meant to let developers move quickly from experimentation to production-ready AI agents. Kong is offering an entry-level Hobby tier for free. Not everything is live yet. Sandboxed compute is expected to arrive shortly after the initial launch. Full general availability is anticipated by mid-November 2026, with November 19, 2026 as the target date. From traffic cop to landlord Kong has been around since 2009 and built its name on API management. Volcano pushes the company into new territory. Instead of only routing the traffic, Kong now wants to supply the runtime and compute that production AI workloads run on. The name will ring a bell for some developers. Kong introduced a Volcano SDK in 2025. The new platform is a different animal, focused on operational hosting and governance rather than a toolkit for writing code. The company has the funding to attempt the jump. Kong raised a $175 million Series E in late 2024, which carried a $2 billion valuation. Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .