The deploy crashed three minutes after you closed the tab. Polylane already knows, and it's working out why.
Railway runs your services, databases, and scheduled jobs from one workspace. Polylane graphs the whole workspace and watches every deploy as it lands.
Join the waitlistEvery supported resource.
The Railway resource types that show up in your graph.
Project Environment Service Deployment Volume Domain TCP Proxy Notification Rule Every deploy gets a second pair of eyes.
Agents graph the workspace, watch its deploys and resource usage, and investigate what breaks.
The whole workspace in one graph
Projects, environments, services, deployments, volumes, domains, and TCP proxies, with their edges and history. Template databases sit next to the services that use them.
Deploys watched as they land
Railway webhooks tell Polylane the moment a deployment fails or crashes. The alert resolves on the next success, and an agent can investigate every deploy if you want it to.
Metrics judged against the service's own normal
Agents comb each service's CPU, memory, and disk against how that service usually behaves. No thresholds to tune.
Logs pulled mid-investigation
Deploy logs and runtime logs are read while the agent works, so the answer arrives with the lines that prove it.
Variable names, never values
Polylane syncs variable names to link services across clouds and never reads the values. The OAuth token is encrypted before it's stored, and the agent never sees it.
Read scopes, and fixes as pull requests
Connect with OAuth using read scopes only. Code changes land as pull requests on the repo behind the service, and any write action pauses for your approval first.
Questions.
Is Polylane an AI SRE for Railway?
It's what most people mean by the term: agents that keep your projects, services, deployments and volumes in one context graph, watch every deploy as it lands, investigate issues and fix what they can, with a human approving anything that changes production. We call it self-operating software rather than an AI SRE: the point is that the project looks after itself, not that a bot joins the on-call rotation.
How does Polylane connect to my Railway workspace?
With OAuth, using read scopes only. Click connect, approve in Railway, and syncing starts. The token is encrypted before it's stored, and the agent never sees it.
What Railway resources does Polylane sync?
Projects, environments, services, deployments, volumes, domains, TCP proxies, and notification rules, each with its edges and history in the graph. The full list is on this page.
How does Polylane know a deploy failed?
Railway webhooks alert on failed and crashed deployments the moment they happen. The alert resolves on the next successful deploy, and you can have an agent investigate every deploy rather than only the failures.
Does Polylane read my environment variables?
It syncs variable names, so a service pointing at a database in another cloud becomes an edge in the graph. It never reads the values.
Can Polylane change my Railway services?
Agents investigate and explain. Code changes only ever land as pull requests on the repo behind the service, and any write action, such as a restart or a redeploy, pauses for your approval first, with the exact request and reason.
Does it work with my observability provider?
Yes. Datadog, Sentry, Honeycomb, Axiom, and Grafana Cloud join the graph as first-class sources, and their alerts get triaged by an agent the moment they fire.
What does it cost?
Pricing is public and self-serve: a free plan, paid plans from $80 a month, and pay-as-you-go overage past the included credits. Every plan and limit is published at polylane.com/pricing.