Author workflows and web scrapers with one polyglot SDK. Push to a repo and we build, version, and run every execution — with automatic retries and zero infrastructure.
Scaffold a project, author your workflow with the SDK, and connect the GitHub repo. DagFlows clones the commit, builds and versions it, then runs every node in its own microVM. Watch the whole path, step by step.
Everything between your code and production — building, deploying, versioning, retries, and scaling — handled by one platform and one SDK, in the language you already use.
Author in any language with one coherent SDK, deploy in a click, and let versioning, retries, and per-node isolation keep every run on track.
Author workflows and scrapers in Python, Node, or Go — one workflow model, whichever language you reach for.
from dagflows.authoring import Workflow, Retry wf = Workflow("orders") @wf.node(retry=Retry(max_attempts=3)) def fetch(ctx, inputs): return {"orders": load()} @wf.node(depends=[fetch]) def total(ctx, inputs): ...
Declare each node and what it depends on. DagFlows resolves the graph and runs it in order — fanning out in parallel wherever it can.
Per-node retry policies with backoff — a failed attempt reruns and settles on its own.
Ship any commit or tag — we clone, build, and promote it.
Every deploy is an immutable, addressable version of your workflow.
Every node runs in its own microVM — sandboxed, resource-capped, and torn down when it finishes. No shared state, no noisy neighbours.
One unit, one price. You buy credits and spend them on compute — 50,000 credits per dollar. A run costs the CPU or memory it holds, whichever is larger, for as long as it runs. No per-request fees, and no bundling CPU to memory.
50,000 credits per $1 · pay only for what runs · run estimates at ¼ vCPU · 1 GB · ~1s.
No servers to provision, no pipeline to wire. Connect a repo and your first run is minutes away.
Sign up and link GitHub so DagFlows can reach your repositories — public or private.
github → dagflows Create a workflow or scraper backed by a repo and branch, then deploy a commit or tag.
acme/repo · main Run on demand or add a cron, webhook, or event trigger. We build, scale, and execute.
deploy → run Join the waitlist and run your first workflow or scraper straight from a repo — no servers, no CI, no schedulers to babysit.