What is helipod?

helipod is an open-source, reactive backend you host yourself. You write your backend as regular TypeScript functions, and helipod keeps every user's screen up to date automatically.

Think of a Google Doc. When one person types, everyone else sees the change right away, with no refresh button. helipod gives your whole app that quality, for any data, and you barely have to think about it.

The idea in one minute

You write two kinds of functions. A query reads data. A mutation writes it. Your app subscribes to a query, like "the messages in this chat."

When someone sends a message, they call a mutation. helipod notices the mutation changed data your query was reading, re-runs the query, and pushes the new result to everyone watching that chat. You never write polling, WebSocket code, or cache-invalidation logic. That part is the whole product.

The developer experience is built entirely around that reactive model, with one defining trait: you run helipod yourself, on your own infrastructure, with no managed cloud in the middle.

You write your backend as functions

Your functions live in a helipod/ folder:

Here's a query that lists messages and a mutation that adds one:

import { v } from "@helipod/values";
import { query, mutation } from "./_generated/server";

// Reads data. Your app subscribes to this and gets live updates.
export const list = query({
  handler: (ctx) => ctx.db.query("messages", "by_creation").collect(),
});

// Writes data. Runs as one transaction.
export const send = mutation({
  args: { author: v.string(), body: v.string() },
  handler: (ctx, args) => ctx.db.insert("messages", args),
});

There are three function types in total:

• Query reads data. Your app subscribes to queries and they update on their own.
• Mutation writes data. Each one runs as a single transaction: it all saves, or none of it does.
• Action is for everything else, like calling another API or sending an email. Actions can't read or write the database directly; they call your queries and mutations to do that.

You'll meet each one properly in Queries, Mutations, and Actions. If you want to know how the live updates actually work, that's How it works.

What comes with it

helipod is a full backend, not just a database. Everything here is built and ready to use today:

• File storage for uploads, images, and downloads.
• Authentication: passwords, magic links, social login (Google, GitHub, and more), two-factor, and passkeys.
• Authorization for roles and permissions.
• Notifications by email, SMS, in-app inbox, and push.
• Scheduled functions and crons for work that runs later or on a timer.
• Workflows for durable, multi-step processes.
• Optimistic updates and offline support on the client.
• A dashboard to browse your data, read logs, and run functions.

Auth, notifications, and the rest are components: you turn on only the ones you need, in one config file. See Components.

Run it anywhere

The same app runs, unchanged, in all of these places:

• On your machine with helipod dev.
• On a server with Docker.
• As a single downloadable binary.
• On Cloudflare.

It stores data in SQLite by default, with no setup. When you want a managed database, point it at Postgres with one flag and change nothing else.

How it compares

If you know these tools, here's the short version:

For numbers, Performance has measured benchmarks, with the caveats spelled out.

A few honest limits

• No full-text or vector search yet. It's not built.
• Functions run in-process, so don't run untrusted code from different tenants on one shared deployment.
• No built-in HTTPS. Put a proxy like Caddy or nginx in front for that.
• Multi-node scaling exists but isn't free-forever the way single-node self-hosting is.

Where to go next