Several people (thanks to @mitsuhiko and @samuel_colvin) noted early on that DBOS had too many dependencies. In the latest Python release, we trimmed it down to just 6 direct deps while still implementing durable workflows and queues in one library. We kept: - sqlalchemy +

@pranaygp I remember that you're also working on workflow versioning recently. Curious to hear your thoughts on this.

I used to be skeptical about workflow patching, mostly because of all the horror stories I heard from other developers. But the way DBOS implements it is nice: a patch is just another workflow step. That tiny shift makes the patching interface more intuitive than I expected.

Just released a new feature–workflow patching! Patching lets you safely upgrade workflow code even when you have active workflows. Essentially, you insert a “patch” into your workflow that instructs new workflows to run your new updated code while old workflows safely resume

Excited for this!

Durable execution frameworks are becoming an important infra pattern for agents. Traditional services run in milliseconds. If they crash, you retry. But agents run longer, retries burn tokens, and crash recovery actually matters. @DBOS_Inc is a great example of this emerging

Do you like databases? Do you want to hear two database professors rant about them? Do you need one of those professors to have a Turing Award for databases? If yes, then join Mike Stonebraker and I next Wed Dec 10 @ 1:00pm EST for database hot takes: https://t.co/JXGHtesZzC

I'm so happy to see our users building cool systems on top of DBOS 😎

The biggest challenge with building AI agents is that agents do weird things. They call the wrong tool, or the right tool with the wrong input, or generate a bizarre text output. Best case scenario they crash and fail, worst case they take a wildly inappropriate action. So what

I just saw a LinkedIn ad selling the NeurIPS attendee list. Ah yes, the cutting edge of AI research... and lead generation.

TBT (long overdue post). I really enjoyed #QConSF this year, and it has now become one of my favorite conferences. Thanks to the organizers for putting together a program with so many deep, practical systems talks, and to everyone who stopped by to discuss reliability and

Here is a nice summary of our talk:

TBT (long overdue post). I really enjoyed #QConSF this year, and it has now become one of my favorite conferences. Thanks to the organizers for putting together a program with so many deep, practical systems talks, and to everyone who stopped by to discuss reliability and

In the upcoming webinar, we’re hosting a conversation with two leaders in the database world, Mike Stonebraker and Andy Pavlo. They will discuss the trends most likely to influence the software you build and the architectures you choose. 🧵 🗓️ Weds, Dec 10, 2025 | 1 PM EST

Happy Thanksgiving! Our dev is already shipping code before dinner.

AI apps aren’t simple LLM calls, they’re workflows with nested loops, heavy document processing, and failure risks. @petereliaskraft walks through a RAG pipeline to show why reliability breaks down at scale, plus how durable workflows use database-backed checkpoints to keep

In this video, Alex shows an awesome demo of using DBOS's fork to debug agents (yes, we can fork out your bugs 🍴). Check it out!

Huge thanks to our organizers: @__vishwanath__ @AlexMillerDB @YingjunWu Oussama Saoudi, Benjamin Owad

Yesterday was our final South Bay Systems talk event of 2025 (what a year!). Huge thanks to @julianhyde and @yytian for their awesome talks! Both discussed the challenges of querying data efficiently across heterogeneous sources. Morel is a new high-level language for working

Thanks so much for hosting the podcast! Really enjoyed our conversation.

Gunnar is such an incredible writer! He explains durable execution better than I do. I love seeing more lightweight, database-backed approaches getting attention, and really appreciate the thoughtful shout-out for @DBOS_Inc . Always grateful for folks who push the space forward.