Creating agent workflows and architecting the logic is one thing, making them durable and fail-safe is another👇
New integration for durable agent workflows with @dbos.dev execution - Make sure your agents survive crashes, restarts, and errors without writing any checkpoint code.
Great to see a concrete comparison! If Durable Functions are limited to a single Lambda, does that also mean parallel execution is effectively constrained within that same Lambda? IIUC, fanning out would require either invoking other Lambdas from within steps, or falling back to Step Functions.
We're starting a new series highlighting the features and integrations our team has been shipping each
month.
We shipped Kotlin support, workflow patching & lineage tracking, dynamic cron, configurable alerting, a new MCP server, and more.
Read the full update here:
www.dbos.dev/blog/dbos-ne...
Coding agents are already great at writing durable workflows. But operating them, especially when things break, is much harder.
If multiple workflows fail with unexpected errors, how do you find the root cause, test, and apply a fix?
That's why we built DBOS MCP www.dbos.dev/blog/mcp-age...
Cool new feature: Dynamic workflow schedules!
The idea is to store workflow schedules in Postgres so you can dynamically create, view, update, and delete them at runtime, as well as (coming soon) view them from a dashboard. Plus, our new design supports backfilling and manual triggering.
I wrote a new blog post with @pydantic.dev! With the Pydantic AI + DBOS integration, agent runs are automatically checkpointed to Postgres. Executions are durable and observable with DBOS + Pydantic Logfire.
Check it out: pydantic.dev/articles/pyd...
Hope they record this one!
A presentation so nice, I'm doing it twice.
I'll be delivering my Building Reliable Workflows in Java talk again on Feb 24th, this time in Seattle. If you're a Java (or Kotlin!) developer, come hear all about what you can do with DBOS.
www.meetup.com/seajug/event...
Open source repo: github.com/dbos-inc/dbo...
DBOS docs: docs.dbos.dev/integrations...
This integration lets agents:
- Recover automatically from restarts, failures, and redeploys
- Run long-lived workflows, including human-in-the-loop and parallel steps
- Orchestrate multiple agents reliably
- Autosave progress for auditing and observability
- Manage workflows (cancel, resume, fork)
OpenAI Agents SDK + @dbos.dev integration is now live! You can run OpenAI agents with durable execution backed directly by your database. DBOS embeds as a library, so it works anywhere, even inside a local coding agent.
Check it out, and any feedback is welcome: openai.github.io/openai-agent...
DBOS is coming to skills.sh
You can now teach your coding agent (Claude Code, Codex, Antigravity, anything) how to write durable workflows by just running:
`npx skills add dbos-inc/agent-skills`
Available in Python and TypeScript.
Check out this new release of Pydantic AI: parallel tool execution in DBOS agents with deterministic replay.
The newly added execution mode runs tool calls in parallel, then emits events in a stable order. This makes agent execution reproducible and crash-recoverable without sacrificing latency.
Love this template. It shows how you can build reliable apps in a serverless environment with DBOS.
I had a blast at @pjug.bsky.social tonight. Thanks for having me!
Fivetran - DBOS webinar tomorrow, Jan 28 at 9AM PT/12PM ET
Learn how to build a control plane that safely manages Fivetran Connector SDK state changes using DBOS durable workflows. It enables idempotency, automatic recovery, and enterprise-grade reliability‚ without complex custom orchestration.
This is helpful to deterministically fan-out step execution in parallel within a workflow.
Nobody wants to meet on Friday afternoon, which makes it the best time to focus and code. Happy weekend!
Join our first user group virtual meetup in 2026. Lots of interesting new features we're releasing!
This is a must-read annual review on database systems. 👀
I think for Jan 2026 you may consider Andy's latest blog post. www.cs.cmu.edu/~pavlo/blog/...
Our first meetup of the year! 🍻
Through conferences, meetups, and online platforms, I feel incredibly lucky to have connected with so many of you this year. I learned something from every conversation, and I hope the new year brings even more chances to connect.
Outside of work, I co-organized South Bay Systems with a group of brilliant engineers. We hosted a wide range of talks as well as coffee chats. I'm deeply grateful to the companies that opened their doors and sponsored us. Good engineering ideas deserve to be shared beyond company boundaries.
Along the way, I spent time contributing to several open-source projects. Working with other communities taught me a lot about how to manage open source well. I remember spending several Friday nights helping users debug issues in other projects. I believe good work doesn't stop at the day job.
At work, we started monthly user groups and invited our users to present their work. Hearing how DBOS made people's lives easier always made me smile. Embracing criticism was just as important, because it helped shape the product into something more lightweight and robust.
The keyword of 2025 for me is: community. This includes @dbos.dev, @southbaysystems.xyz, and several open-source projects. 🧵
Thanks to everyone who was part of my 2025. I'm proud of what we built, clearer about what matters, and excited about the year ahead.
DBOS 2025 Highlights 🎄 DBOS Transact (Open-source durable workflows library) Durable Queues: priority, rate limit, concurrency, debounce, deduplication, dynamic partitioning, heterogeneous workers Workflow Management: timeouts, cancel, resume, fork Versioning & Patching: seamlessly upgrade your code DBOS Client: interact with DBOS apps from other services TypeScript: support more runtimes including Bun and Deno Python: support SQLite; sync and async functions New Languages: Go, Java DBOS Conductor (new!) (Out-of-band management UI) Distributed failure recovery, data retention, observability On-prem self-hostable, privacy preserving Partnerships and Integrations Supabase, Vercel, Neon, Tiger Data, Pydantic AI + Logfire, PlanetScale, Datadog, … more to come!
2025 has been a wild year for DBOS. Huge thanks to you all for the support, feedback, and trust along the way.
Looking forward to making reliable software even easier to build in 2026 🚀
Simplicity wins. I really appreciated the shoutout: "Integrating durable workflows through DBOS solved our distributed system headaches without adding ops overhead."
www.decodingai.com/p/building-v...
