Launched this week

Volcano
The self-healing agent cloud by Kong
83 followers
The self-healing agent cloud by Kong
83 followers
Volcano is one place to build, host and run agents and apps. Deploy from the coding agent you already use (Claude Code, Codex, Cursor, OpenCode, Pi) or the AI Builder. You get year-long durable functions, human-in-the-loop approvals, resumable sandboxes, distributed locks, branchable Postgres, auth, storage, realtime, and backend and Next.js hosting. LavaFlow finds and fixes bugs on its own, so your apps and agents keep getting better over time.






Free Options
Launch Team

Memanto CloudYour agents focus. Memanto remembers.
Promoted
Hey Product Hunt, we're the team at Kong behind Volcano.
Building an agent today usually means wiring up a separate service for backend hosting, Next.js hosting, the database, user management, file storage, realtime, workflows, approvals, sandboxes and locks, then writing the glue code between them. When something breaks, it waits for a person to find the issue and ship a fix. We kept running into that building agents for ourselves at Kong, so we put it all in one place.
Volcano is a self-healing agent cloud. Build with the coding agent you already use (Claude Code, Codex, Cursor, OpenCode, Pi) or the AI Builder, and deploy to one project that has:
Agent primitives: year-long durable functions, human-in-the-loop approvals, resumable sandboxes and self-renewing distributed locks
Hosting: serverless backends and CDN-accelerated Next.js
App backend: branchable Postgres, user management, private-by-default file storage and realtime
Then LavaFlow watches what's running, finds bugs and writes the fixes on its own, so your apps and agents keep getting better over time.
You can start on the free plan at volcano.dev. Tell us what you'd build, and what's missing.
Is Sydney one of the regions you can pick on the Superagent plan?
does LavaFlow ask before shipping a fix?
@leoheo by default it opens a PR and stops, nothing ships until you merge.
@connor_church phew, i get to keep the merge button then lol
Deploying straight from the coding agent I already use is the part that caught my eye — I live in Claude Code most days, so not having to context-switch to a separate deploy flow matters a lot. The year-long durable functions plus human-in-the-loop approvals feels like the right combo too: long-running agents need to be able to pause and wait on me, not just run until something breaks. How do the resumable sandboxes handle a task that sits waiting on approval for days — does it pick up exactly where it left off?