Durable execution is underrated. In real systems, any step can fail at any time. Peter breaks down why reliability is such a challenge.

Building reliable AI systems is hard. In this video, @petereliaskraft.net walks through a deep research agent that many of us want to build and the kind that fails in all the surprising ways. Failures are inevitable and a durable workflow is what will make your agent production ready.

When Yutori looked into Temporal, the cost was clear. It would mean rearchitecting their workflows. With DBOS, they could orchestrate durable workflows using just a few Python decorators and a Postgres database.

Celery’s familiar. But once tuning and load management start piling up, things get messy fast. DBOS changes that.

If you’ve ever wondered why your queue is filling up, this one’s for you. Cameron shares how DBOS’s Grafana dashboard helps them see every job, every queue, and fix stuck workflows before things go sideways.

When reliability is built in, devs can focus on solving real problems instead of managing infra. Hear what Cameron Spiller from Soria Analytics has to say about it.

What’s the real impact of switching to DBOS? Simpler workflows, better observability, lower infra costs. Hear how it transformed productivity for Dosu.

As LLMs and agents become the foundation of modern software, every stack needs durability and state recovery. DBOS brings that to any agent stack, without the complexity.

Hear what Devin from Dosu has to say.

Versioning in DBOS Cloud makes deployments clear and reliable, here’s how Yutori is using it:

✅ Every new release spins up its own VM
✅ Logs are tagged by version, so you know which release produced them
✅ Full observability across versions without losing context

When Yutori was scaling their multi-agent framework, they faced a choice: rewrite everything, or find a simpler path.

DBOS made that decision easy. Instead of heavy rewrites, Yutori just added decorators. Instead of new infrastructure, they kept Postgres.

Reliable software starts with durability. @qianli.dev's research at Stanford revealed how to give code the same guarantees as databases.

As @jedberg.bsky.social put it on Last Week in AWS with @quinnypig.com, moving data reliably in agentic workloads for AI training and inference is tough. DBOS makes it simple.

Congrats @qianli.dev on being selected as Silicon Valley Business Journal 40 Under 40 honoree 🥂

Struggling to deploy Lambda functions from development into production? You're not alone.