# One platform for your app and its database (/compute)

> Prisma Compute | Deploy TypeScript Apps and AI Agents on Bun. For the complete Prisma website index, see [llms.txt](https://www.prisma.io/llms.txt). A Markdown version of any listed page is available at its `.md` URL — this one is `/compute.md`.

Prisma Compute deploys TypeScript apps, APIs, and AI agents from your repo as long-lived Bun processes next to Prisma Postgres, with long-running requests and streaming. One platform for your app and its database.

Prisma Compute is the easiest way to host TypeScript apps, with Postgres and the rest of your infrastructure built in.

- [Get started free](https://console.prisma.io/sign-up)
- [Read the docs](/docs)

Deploy from GitHub or the CLI. Free tier, no credit card.

- Every PR gets a dedicated app and database, branched together
- Microsecond app-to-database latency, no cross-vendor network hop
- Long-running, streaming workloads built for hosting agents, not just apps

### Prisma Compute at a glance

- **Connect** — Point Compute at a GitHub repo, or deploy straight from your machine with one command.
- **Deploy** — One deploy ships the app and its branched database together, migrations included.
- **Live** — Production and a per-PR preview, each with its own app and its own isolated database.
- **Apps** — Start from an app — Next.js, Hono, an API, an agent — and deploy it as it is.

## One platform can ship features two never could

When your hosting provider and your database provider are separate companies, the things that depend on them being aware of each other (branching, preview environments, low-latency queries, end-to-end test data) don't exist. You wire them together yourself, or you do without.

Prisma Compute exists to change that.

- Branch your stack per PR
- Skip the cross-vendor network hop
- Host streaming agents natively
- One config for app and database

## What one platform unlocks

Compute and Prisma Postgres run as a single platform. The features below are only possible because of that.

- **Built for hosting AI agents** — Agents need long-lived processes, streaming responses and durable memory. Compute supports all three, on the same platform that runs your data.
- **Branch your stack per PR** — Every preview environment gets a dedicated database, branched from production and tied to that deploy. No shared test DB that passes tests production fails.
- **Microsecond queries, no code changes** — ORMs make many small queries, and on most stacks each one pays a network round trip between two vendors. Compute runs on the same machine as Prisma Postgres, so that round trip disappears.
- **Cron and background jobs as a first-class concept** — Scheduled execution and background jobs are declared in prisma.config.ts, alongside the rest of your app. No external schedulers, no separate sidecar service.
- **[S3-compatible file storage built in](/docs/compute/object-storage)** — Object Store buckets live inside your project, next to its databases, managed from the same Console and API. Any S3 client works, with per-bucket keys scoped to read or read-write.

## Built to work with the rest of Prisma

Prisma Compute is where the stack comes together: your app, your database, and your data model on one platform.

**Best together. Swappable when needed.**

## One platform. Both layers. Features only possible because of it.

Compute is part of a TypeScript stack where hosting and the database travel as a unit. Branch them together, deploy them together, debug them together.

- Built to host the agents you're building, not just the apps
- Per-PR branching across app and database
- Co-located queries with no cross-vendor latency

- [Get started free](https://console.prisma.io/sign-up)
- [Read the docs](/docs)