# Your agents ship faster than your CI.

Agents multiplied your commits, and CI became the wall. Ravion pipelines build, test, plan Terraform, and deploy on Firecracker sandboxes with whole-disk caching, on any instance size you choose. Every run starts warm from your last build, and your code, caches, and logs stay in your own AWS account.

[Start free](https://app.ravion.com/signup)

[Read the docs](https://www.ravion.com/docs/concepts/pipelines)

- **sandbox boot:** ~1 s
- **warm rebuild with whole-disk caching:** ~5 s
- **per minute after 50k included each month:** $0.002

Trusted by engineering & platform teams at scale

## CI was built for humans who push a few times a day.

Agents push hundreds of commits a day. Every one waits in line for a runner, reinstalls the same dependencies, and rebuilds from scratch. The agent finished the fix in four minutes. Your CI is still on step two.

### Production keys in someone else's cloud

Deploys need AWS access, so you hand a CI vendor the keys to production.

### Every build starts cold

Fresh machines reinstall the same dependencies and rebuild the same layers. Every time.

### Context everywhere

CI logs in one tool, infrastructure in another, deploys in a third.

With Ravion

## Builds that remember.

Sandbox steps keep their disk between runs, so builds pick up where the last green run left off.

Step 1

### Run every step on a warm sandbox

Firecracker microVMs boot in about a second. Add disk\_cache: auto and each run restores the whole disk from the last green build, with packages downloaded and Docker layers cached.

Replaces GitHub Actions, CircleCI, Buildkite

[See step types](https://www.ravion.com/docs/pipelines/step-types)

Step 2

### Build, approve, deploy

Images, static sites, custom commands, approvals, and deploys. In parallel when you need it.

Replaces separate CI and CD tools

Step 3

### Terraform in the same run

Plan and apply are just steps. Infrastructure and code ship together, with one approval.

Replaces separate Terraform CI

Step 4

### One pipeline, every environment

Define it once, add a variant per environment.

Replaces copy-pasted workflow files

[Learn about variants](https://www.ravion.com/docs/pipelines/variants)

## Your pipelines. Your cloud.

Ravion orchestrates. Your account does the work.

Interact with Ravion through the UI, CLI, API, or Git. Ravion sends steps to your AWS account and gets status back. Everything else runs inside your account.

-   UI
-   CLI
-   API
-   Git

RavionControl plane

-   Orchestrates steps and approvals
-   Variants per environment
-   Logs and history

Steps

Status

Your AWS accountYou own it

Your VPCPrivate network

Firecracker sandboxes and EC2Any size

-   build
-   test
-   plan
-   deploy

Builds, tests, and deploys run here

Private services

-   Databases
-   Internal APIs

No allowlists

-   Cached disks in S3
-   Your IAM and AWS bill

## Fast by default.

### One-second microVMs.

Every sandbox step gets its own Firecracker microVM, booted in about a second.

### Ultra-fast whole-disk caching.

Every run starts from the whole disk of your last green build, so packages and Docker layers are already there.

### Branch-aware.

Pull requests start from their base branch's disk.

### Spot pricing.

Steps that can safely rerun use Spot capacity.

### Runs in your VPC.

Reach private databases without opening anything up.

### Built for agents in the loop.

Your agent can start a pipeline, wait on it, read the step logs, and retry it from the Ravion CLI or API, with no tab-watching and no polling GitHub.

## Minutes on machines you own

You pay AWS for compute and Ravion for minutes.

TodayWhat it costs youWith Ravion

Hosted CICircleCI / GitHub Actions runners / Buildkite

What it costs you

$0.012/min on larger GitHub runners, plus per-seat fees on some plans

With Ravion

50k min/mo included, then $0.002/min, on machines in your account

## Bring your workflows with you.

Your commands and Dockerfiles move over as steps. Run both side by side until the results match.

Teams move their pipelines off GitHub Actions when CI becomes the bottleneck.

-   Slow, fragile hosted runners
-   Caches rebuilt from scratch every run
-   Minutes that cost far more than the compute
-   Secrets and source code running on shared infrastructure

GitHub ActionsRavion

Workflow filePipeline config

JobStep

on: push and pull\_requestGitHub trigger

Deployment environmentVariant

runs-onStep infrastructure

Move in three steps

[Read the config reference](https://www.ravion.com/docs/pipelines/config-reference)

1.  01

    Turn each job into a pipeline step and each deploy target into a variant.

2.  02

    Add a GitHub trigger for push or pull\_request events.

3.  03

    Run both side by side until the results match, then delete the workflow.

Pipeline questions

### Do we have to replace GitHub Actions?

No. Keep your CI and trigger Ravion deploys from it. Or run your GitHub Actions jobs on Ravion runners. Or move everything into Ravion pipelines. [CI integration](https://www.ravion.com/docs/config-as-code/ci-integration)

### How does whole-disk caching work?

A sandbox step with disk\_cache keeps its disk. The next run restores the last green run's disk from its branch, then its base branch, then the default branch. Only green runs publish.

### What can trigger a pipeline?

Git pushes, the API and CLI, the dashboard, and other pipelines.

### Where does the config live?

In your repo, as YAML, reviewed like any other code. [Pipeline config reference](https://www.ravion.com/docs/pipelines/config-reference)

## Ship code and infrastructure in one run.

Run your first pipeline in minutes.

[Start free](https://app.ravion.com/signup)

[Book a demo](https://www.ravion.com/demo)

Free to start / No credit card / Runs in your AWS account