Survives deploys
Every step is checkpointed. A rolling deploy mid-job is a non-event.
01$ Durable execution for background jobs
$ asgardbyte --init
Retries, replays and exactly-once steps as a primitive — jobs that survive deploys.
# install npx asgardbyte init --policy default ✓ connected · 0 duplicate charges in 18 months ✓ first run recorded
Overview
Runs
Policies
Traces
Settings
Durablelive
7Mjobs replayed after a failover
0duplicate charges in 18 months
Survives deploysok
Deterministic replayok
Exactly-once side effectsok
Teams shipping with Asgardbyte
NorthwindKestrelRuangPolarTokoLumenBakauVela
What you get
Three things Asgardbyte does properly.
Deterministic replay
Re-run a failed job from the exact step, with the same inputs and the same clock.
02Exactly-once side effects
Idempotency keys are generated for you, not remembered by you.
037Mjobs replayed after a failover
0duplicate charges in 18 months
11 saverage recovery time
Inside the product
One screen, the whole picture.
Overview
Runs
Policies
Traces
Settings
Durablelive
7Mjobs replayed after a failover
0duplicate charges in 18 months
Survives deploysok
Deterministic replayok
Exactly-once side effectsok
“We stopped writing compensation logic. The workflow just resumes where the pod died.”
Jonas K. — infra engineer, Berlin
Pricing
Start free. Pay when it earns it.
Questions
The things people ask first.
Is it a queue?+
No. A queue moves messages; this keeps state. You still keep your broker.
What happens on a poison message?+
It parks with full context, and the workflow continues on the branch that can proceed.
Can I self-host?+
Yes, single binary plus Postgres. Same API.
Asgardbyte is open for design partners.
Small team, direct access to the people who build it. Bring a real workload.
Request access