🎙️ How does durable execution make AI more reliable? Ryan Donovan is joined on the podcast by Jeremy Edberg, CEO of DBOS, and Qian Li, co-founder, to dive into its use cases, and how it's changing the future of debugging and observability for tech. stackoverflow.blog/2025/05/20/dur…

One subtle challenge in building durable workflows is how to handle different code versions. Durable workflows work by checkpointing workflow state (which steps completed and what their outcomes are) in a database, then using that checkpointed state to recover from failures.The

A couple users recently asked me how to schedule workflows to run sometime in the future. The good news is that DBOS makes this really easy, even if you want a workflow to run days or weeks from now. The key is durable sleep. If you need a workflow to wait an hour, day, or week

How did Google build the first database with truly global scale? Spanner is an absolute classic, one of the most interesting papers I’ve read and one I keep coming back to. It presents an incredibly impressive system: a globally-distributed database running at the largest

Spanner is a great example of the use of synchronized clocks in distributed systems. The clock API is critical for global consistency, concurrency control, transactional guarantees, and high performance.

Building reliable systems is a continuous process of introspection and iteration. This week, the DBOS team gathered to review what we've learned and plan for the next steps. We're dedicated to making durable workflows and queues lightweight so that developers can adopt them with

Congratulations Yutori on the launch! As a startup founder, few moments compare to watching your customers succeed and knowing your product played a part in their journey. Yutori has been an amazing design partner, using DBOS to build the workflows powering scouts. They’re

New feature: Adjustable retention policies for your DBOS applications! You can now configure for how long applications retain workflow history, either by time or by the number of workflows you want to retain. This is useful for compliance or to keep database storage usage in

The most underrated benefit of durable workflows is observability. Want to watch your app’s progress in real time? Pinpoint exactly which step your data pipeline failed on? Check how long each workflow takes to run? Monitor the progress of each of your queues? Normally, this

Join us for the June DBOS User Group Meetup next week! We're excited to welcome the engineering team from Yutori, one of our customers building powerful AI agents that can autonomously perform digital tasks across the web. They'll share how they're using DBOS to make

Want to build AI applications that don't break in production? Join us next week at the virtual MLOps Summit, where Elliot Gunton, Alan Nichol , and I will share lessons from building AI applications that can withstand the messiness of the real world: where APIs fail, users take

Just listened to a fascinating talk from Yutori about operating AI agents at scale. Yutori has a bold vision: building an AI chief-of-staff that can reliably handle everyday digital tasks. They've just launched "scouts,” which are self-directed research agents that

To build reliable systems at scale, you need timeouts. By preventing operations from holding resources indefinitely, they make applications far more resilient to failures, deadlocks, and performance bugs. However, timeouts are tricky to implement correctly. To be robust, they

One of the most interesting things about agentic workflows is that they’re unpredictable. Normally, a business workflow follows a well-defined execution graph. For example, a checkout workflow might say "if payment succeeds, then fulfill the order, else ask the customer to try

What if your database could have ALL the memory? When you’re running a database at scale, the most expensive part is memory. Databases primarily use memory for their buffer pool, a local cache of data objects stored on disk. Because the buffer pool is a cache, databases don’t

CXL technology is promising for memory disaggregation in databases. This paper explores CXL for memory disaggregation in databases. Great to see how hardware advancements enable simpler software design and better performance. A key insight from the paper is that directly running