Push a repo. Get your whole stack running.
Avalon reads your repository, works out what it actually needs — the database, the cache, the queue — provisions every piece on our managed Kubernetes clusters, and ships each service with TLS and its connection strings already wired in. You watch the whole thing happen, line by line. Right now it costs nothing.
Three steps, and then it keeps doing it on every push.
No Helm charts to write, no Terraform to maintain, no YAML to copy between environments — and for the free beta, no cloud account either. Connect a repo and Avalon handles the rest.
Connect a repository
Connect your GitHub account, then pick a repo and a branch. No cloud account to link, no cluster to stand up, nothing to install.
Avalon works out what you need
It clones, reads the stack, and finds every backing service your code talks to — then provisions each one and injects the connection strings as environment variables.
Ship it, and watch it ship
Split your quota across services however you like, pick public or private, then deploy. Every event streams back live. Turn on auto-deploy and each push to the branch does it again.
100 cores and 200 GB of RAM. We're giving it away.
We put a pool of our own cluster capacity aside to find out what people actually build on Avalon. Every account gets a slice of it — free, no card, no trial countdown in the corner of the screen. It runs on our managed clusters, so there is nothing for you to provision.
2000 millicores, yours to split across as many services as your app needs.
Enough for an API, a worker, a Postgres and a Redis with room left over.
Provisioned dependencies, TLS, live logs, auto-deploy on push. No feature gates.
Export or delete it whenever you want. Nothing is held hostage to a plan.
What happens when the pool runs out. This is a few days of free capacity while we learn what people deploy — not a permanent free tier we're pretending to fund. You'll hear from us well before anything is switched off, and nothing gets deleted without warning. Pricing comes later, once we know what we're actually worth to you.
| Avalon | Frontend hosts | Rolling your own | |
|---|---|---|---|
| Deploys your app | Yes | Yes | After you build the pipeline |
| Provisions databases & queues | Detected and provisioned | Bring your own | You write the Terraform |
| Runs your backend services | Any container | Functions only | Yes |
| Kubernetes to maintain | None | Not applicable | All of it |
| Cost to start | Nothing | Free tier, then usage | A cloud bill on day one |
| Time to first deploy | Minutes | Minutes | Weeks |
The platform gets audited. The workload doesn't.
We spent our careers breaking into environments like the ones these platforms produce: flat pod networking, unsigned images, containers running as root. The control plane holds up to the audit; the thing it deployed never gets looked at. Closing that gap is why Avalon exists — and since we run the clusters, it's ours to close, not yours.
Where we actually are today. Automated deployment is live and running real beta workloads. The hardened defaults below are rolling out over the coming weeks — we'd rather show you the order we're shipping in than imply it's all done.
Automated deployment
Stack detection, dependency provisioning, managed rollouts, live event streaming, auto-deploy on push, per-service resource limits.
Account security
TOTP and email multi-factor auth, recovery codes, scoped API tokens, and TLS on every deployed service.
Secrets and env vars
Per-environment variables with masked values, bulk .env import, and connection strings injected without ever passing through a commit.
Hardened node images
CIS-benchmarked base images for every node we run, rebuilt and re-scanned on a schedule rather than at install time.
Default-deny networking
Pod-to-pod traffic denied unless declared, so a compromised service can't walk sideways through your environment.
Admission control
Unsigned images rejected, root containers rejected, privileged pods rejected — enforced at the cluster, not left to the developer.
Edge protection
WAF, DDoS mitigation and bot filtering in front of every public service, on by default instead of as an upsell.
Automated scanning
Image and dependency scanning wired into the deploy path, with findings attached to the service that introduced them.
Compliance evidence
The posture report your prospects keep asking for, generated from what's actually enforced around your workloads.
Your next deploy could take four minutes.
Create an account, connect a repository, and watch Avalon work out the rest.
- 2 vCPU and 4 GB RAM on our clusters, free
- No credit card, and no trial clock
- 33 of 50 seats still open in the pool