Lakefront ships your services to your own Azure, AWS and GCP accounts, and moves them between regions and providers the moment one goes down.
Works with Azure · AWS · GCP · and your AI agents, via auth.md.
Your services sit behind 300+ edge points of presence and route to the nearest healthy region, across all three providers. When a region drops, requests land somewhere else in milliseconds.
The same definition compiles to each provider’s native primitives. Start on Azure, add AWS next quarter, never rewrite a line. No Kubernetes to babysit, no YAML to memorize.
export default service({ name: 'checkout-api', source: repo('acme/checkout'), cloud: 'azure', // or aws · gcp regions: ['eastus', 'westeurope'], scale: { min: 1, max: 10 }, })
Each agent signs in as itself, asks for only the access it needs, and does the work. It never goes past your permissions, and it stops the moment you revoke it.
Every action is scoped and logged under the agent’s own name, using the openauth.md protocol that Cloudflare and Resend run.
Grant a person or an agent access once. Lakefront maps it to the right IAM roles in each cloud, so you stop reasoning about four permission models at midnight.
Access is brokered per request and expires on its own.
Deploy this service, read those secrets, nothing wider.
Pull a grant and it’s gone from every provider at once.
A single audit trail across Azure, AWS and GCP.
Connect a repo and every push builds an immutable image, runs your tests, and rolls out behind a health check. The CI/CD is already wired. You don’t author it.
Each deploy is a clean build in your own registry.
Traffic shifts only once the new revision passes.
Any past revision is one click away, on any cloud.
Ship a real preview environment for every branch.
Lakefront is a control plane, not a host. Your services, data and bill stay inside the cloud accounts you already have, switch us off and everything keeps running.
Free while you build. You only ever pay your own cloud.