NOV 17 · AI Infrastructure Night · SF

Infrastructure for software outside your cloud.

Deploy and operate software in remote environments - close to sensitive data, private networks, and latency-critical workloads.

Get started Talk to us
ImageControl plane
Deployments
0 connected
Release
3f9a2c1
Rollout
Complete
  • Incredibuild
  • AgentMail
  • Boundary
  • microsandbox
  • Autumn

Deploy to your customer’s cloud without relying on them to operate your software.

ImageImageImageImage

Fix incidents in remote environments without SSH or open ports.

Deploy to remote Linux servers with auto-updates and observability.

$ sudo acme-deploy join

Encrypt your customers’ sensitive data with keys they can revoke.

Run AI inference inside your customer’s VPC.

  • ImageImage Amazon Bedrock
  • Image Vertex AI
  • Image Azure AI Foundry
  • Open-weight on GPUs

Give each customer a dedicated, isolated deployment, in your cloud or theirs.

customer-1 · EUcustomer-2 · India
customer-3 · UScustomer-4 · UK
customer-5 · Singaporecustomer-6 · Brazil

You send them a Helm chart.Then it’s out of your hands.

# ext-customer-12 organizations
Customer 1’s platform team 10:12 AM

Our data can’t leave our cloud.

Your engineer 10:31 AM

No problem. Here’s our Helm chart.

acme-1.8.0.tgz
Thursday
Customer 1’s platform team 2:14 AM

Production is down. Can you get on a call?

Customer 1’s platform team started a callJoin

Alien makes their cloud feel like yours.

  • Auto updates

    Release once, every customer updates.

  • Observability

    Their admin sets telemetry: automatic, approval required, or off.

  • Remote commands

    Fix issues without logging in.

Build a control plane ⇄ data plane architecture, so one breach or one noisy signup can’t hit every customer.

The same architecture behind
  • ImageAWS
  • ImageDatabricks
  • ImageSnowflake
  • ImageCloudflare
  • ImageMongoDB
Learn more
  1. 01

    One shared deployment

    Most SaaS products serve every customer from the same servers, the same cloud quotas, and one database.

    One breach exposes everyone. One noisy signup can slow down your biggest customer.

  2. 02

    Control plane ⇄ data plane

    Some parts only manage the product, things like sign-in, billing, and settings. They don’t need customer data. This is the control plane.

    The parts that process and store customer data are the data plane.

  3. 03

    A data plane per customer single-tenant

    Each customer can then get their own data plane, with its own servers, quotas, and database. This is called single-tenant.

    A breach in one can’t reach the others, and one customer’s traffic can’t slow down another.

  4. 04

    In their cloud BYOC

    A data plane can also run inside the customer’s own cloud account. This is called BYOC, bring your own cloud.

    Their data never leaves their account.

  5. 05

    Small, with no way in

    A data plane often runs where you can’t log in to fix it. That’s why it’s kept small, with few dependencies: less can break.

    Ideally it accepts no connections from the internet. It connects out to the control plane, so there’s nothing for an attacker to reach.

  6. 06

    Every customer

    The hard part is operating all these data planes: deploying, updating, and fixing each one.

    Alien does that for you, in your cloud or in your customers’ clouds.

One shared deployment. Most SaaS products serve every customer from the same servers, the same cloud quotas, and one database.

Wherever your customers’ data
has to stay

  • ImageImageAWS
  • ImageGoogle Cloud
  • ImageAzure
  • On-prem

Sell to customers whose data can’t leave their cloud.

  • Their account, their region

    Data stays where it is.

  • Least privilege

    No access to customer data.

  • Revocable anytime

    They can end access whenever.

  • Updates and telemetry built in

    Every deployment reports in.

Their admin sets it up from your page.With their own cloud credentials. You never get their login.

Deployment UX
  • How it works
AcmeBring your own cloud › AWS
Deployment namecustomer-1
Platform
Configure
Install method
CloudFormationAWS stack launchAWS
  1. Review the stackOptional, but useful before handing this to an AWS admin.Review the CloudFormation template
  2. Choose regionThe AWS Console opens in this region.us-east-1 · N. Virginia
  3. Launch in AWSCloudFormation creates the resources and registers the deployment.ImageImageLaunch in AWS Console

After AWS finishes, return here to watch status and logs.

Ship to remote environments at the speed of light.

fix: retry webhook delivery3f9a2c1 · main
  • ImageImagecustomer-1
  • Imagecustomer-2
  • Imagecustomer-3
  • Imagecustomer-4
  • One command

    From your laptop or CI.

  • Zero credentials

    Alien never holds keys to their cloud.

  • Rollback

    Pin or roll back any deployment.

Build against one set of well-tested primitives.

All resources
  • ImageImageAWS
  • ImageGoogle Cloud
  • ImageAzure
  • ImageKubernetes
  • Your laptop
  • WorkerImageImageLambda
  • StorageImageImageS3
  • QueueImageImageSQS
  • VaultImageImageSSM Parameter Store
$ alien devThe whole stack on your laptop. No cloud account.

Encrypt customer data with a key they control.Give them a cryptographic kill switch for their data.

Customer-managed keys
Your product
Q3 numbersReadable
Hi Bob, the Q3 numbers are attached.Revenue closed at $41.2M, up 18%.Please keep this internal.
Their key
Your customer’s cloud
try it

Run AI models in your customers’ clouds.

Local AI models
Your productone request
POST ai.alien.dev/customer-1/v1/chat/completions{  "model": "byo/claude-opus-5",  "messages": [ … ]}
AI Gateway
  • Customer 11,285 requests todayAmazon Bedrockus-east-1 · billed to them
  • Customer 2913 requests todayVertex AIeurope-west1 · billed to them
  • Customer 3406 requests todayAnthropicbilled to them

Start from a working example.

See all examples
The connector reads credentials from the Vault and queries the database inside the customer's network. Only the rows you allow come back. Read the exampleThe agent reasons in your cloud and sends tool calls as commands. The worker runs them next to the files, and only the result comes back. Read the exampleThe chat app, the database, and the model all run in the customer's account. Only releases come from your side. Read the exampleWebhooks land on the Worker's public endpoint and are stored in the customer's Kv. Your backend queries them with commands. Read the exampleThe agent reads the Kubernetes API and connects out. The Operator, which the chart installs, pulls releases and sends logs. Read the exampleThe dashboard and its metadata stay hosted. The agent and the GitHub token stay in the customer's account. Read the example

Deploy into their cloud. Operate it like yours.

Built in the open. Contribute freely.

alienplatform/alien Slack community