Modal. Serverless GPU apps, functions, and sandboxes.
Connect your Modal workspace to map its environments, apps, functions, web endpoints, sandboxes, volumes, secrets, queues, and dicts into the context graph. Agents can then reason about what runs where, what each function depends on, and what changed between syncs.
Join the waitlistIn the graph.
The Modal resource types Polylane syncs, with their edges, config, and history.
Environment App Function Class Sandbox Image Volume Secret Queue Dict The tools agents get.
What an agent can do with Modal mid-investigation. Every call lands in the thread transcript, cited.
modalGetLogs Get Modal app logs: per-bucket volume by stream plus sampled lines
modalListSandboxes List Modal sandboxes in an environment, with their state and resource shape
modalListDeployments List Modal app deployments: version, who deployed, and when
modalDescribeApp Describe one Modal app: state, deploy provenance, and the functions inside it
modalGetCost Get metered Modal spend, for one app or broken down across the workspace
Questions.
Is Polylane an AI SRE for Modal?
It's what most people mean by the term: agents that keep your Modal resources in one context graph, watch them, 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 your software looks after itself, not that a bot joins the on-call rotation.
Can Polylane change anything in Modal?
Only if you let it. Connected read-only, agents investigate and explain, and changes only ever land as pull requests for you to review. With write access, every write pauses for your approval first, with the method, path and reason on screen. Rollbacks are off by default: you enable them per automation and can turn them off at any time.