On my second day as a Temporalite, I walked into Moscone South not knowing a single person in the building. I joined Temporal's marketing team just days before and immediately found myself at Replay — the company's annual conference, in the heart of San Francisco, surrounded by 2,000 engineers! As someone who had yet to meet my coworkers, I had no idea what to expect.
The badge pickup line was the first test. I passed, mostly because someone handed me a Ziggy stress ball before I could overthink it, and from that moment the nerves were mostly gone. What followed over the next three days changed how I thought about the company I had just joined.
| The crowd in the main lobby during day two of the conference |
Meeting the community#
I was a Software Engineer in my previous role, so I couldn't wait to talk to other engineers in the community using Temporal. The Hackathon was the natural place to start.
It was a great chance to pick the brains of developers about different things, and a big part of it was hearing what they thought about using AI and Temporal in their workflows.
Rene Turcios, an independent contractor, put his finger on a pain point gaining real traction within our community: rate-limiting for AI heavy workflows. As more developers lean on AI to do the heavy lifting in their code development, especially through vibe-coding, current orchestration patterns aren't always built to absorb the load.
Dan Marinescu, formerly of Coin Tracker, brought a grounding case study from his own experience about life without Temporal. Before implementing durable execution solutions, managing workflow pipelines across millions of user wallets meant flying blind when failures happened — redundant error-handling code everywhere led to a lack of visibility into what was going wrong.
Khalid Shaikh of Apple and Rudy Banerjee of Geico raved about the compatibility of Temporal in enterprise software. In their own words, they believe that Temporal is especially well-suited for large scale problems.
| Hackathon group collaborating on their project |
One project stood out above everything else. Anisha Suri and Bashar Ammari of Gilead Sciences built a portal that uses Temporal to monitor long-running clinical trials. As someone who has always believed cutting edge technology has an obligation in healthcare, watching them demo it hit differently than the other projects.
Simultaneously, workshops in Java, Go, and Python went on throughout the day as well. Rooms were packed, with attendees fully engaged with the content while helping the person next to them.
Galvanized by leadership and adoption#
Day two opened with a keynote from Samar Abbas, Temporal's CEO and co-founder. If anyone had walked in uncertain about where Temporal was headed, they didn't leave that way.
| Samar Abbas during the keynote on day two |
Samar then introduced one of Temporal's biggest customers to the stage: OpenAI.
Venkat Venkataramani, VP of Applied Infrastructure at OpenAI, stepped up to share how Temporal has become foundational infrastructure for them. OpenAI started using Temporal in Q1 2025 and grew 60x in their usage under a year, with Temporal now underpinning agentic workflows as well as infrastructure and business processes across the organization.
In Venkat's words: "Temporal scales well. It is one of those systems that actually survives making contact with massive production workflows."
| Venkat Venkataramani during the keynote on day two |
His message was eye-opening to me as a new Temporalite. It made me proud to know the company I'd just joined was at the heart of a tool that so many people — myself included — rely on every day.
Breakout sessions#
The breakout sessions covered more ground than I could fully absorb in a day, but several stories stuck with me.
Netflix's Rob Zienert presented a five-year story about organizational change. Starting with a single Temporal use case hidden inside Spinnaker in 2020, Netflix's adoption grew organically through champions, word of mouth, a pivotal move to Temporal Cloud in 2021, and eventually an announcement on Netflix's internal "Paved Road" in 2024. Today, over 150 teams use Temporal at Netflix across 500+ namespaces, processing more than 7 billion actions per month.
| Rob Zienert answers an attendee's question during a breakout session |
Rob's take was that earned adoption is more durable than mandated adoption. The Paved Road doesn't force engineers to use Temporal. It just makes Temporal the obvious best choice.
NVIDIA's Jeremy Cook and Matt Tescher explained how Temporal powers autonomous vehicles infrastructure. Matt's team built TestStudio, a purpose-built platform that uses Temporal to orchestrate the execution of millions of simulations. Jeremy, who has over five years of Temporal experience and is particularly excited about how physical AI simulation will accelerate complex systems development, made the case that durable execution is a key foundation for safely building the future of transportation.
As he put it: "At the end of the day, we want to move things that we don't have to run to something else. That frees our team to work on business value."
Hearing that from someone building infrastructure for autonomous vehicles hit close to home — freeing engineers from fault tolerance so they can focus on actual product work is one of the key things that drew me to Temporal in the first place.
The breakout sessions led by Temporal's own engineering teams gave me a glimpse into the incredible work being done internally. Johann Schleier-Smith and Ethan Ruhe unveiled new AI capabilities designed to make durable agentic applications simpler and more reliable in production.
| Johann Schleier-Smith speaks on stage at a breakout session |
In another session, Temporal's Carly de Frondeville and Stefan Richter, joined by Sagar Randive from Google Cloud Platform, were announcing major Worker improvements under the banner of "Build with Confidence," giving developers more control, better deployment safety, and a smoother operational experience.
One of the most compelling adoption stories was told by Cursor's Jeremy Stribling. Cursor's cloud agents run on Temporal at serious scale: over 50 million Temporal actions per day, across more than 7 million unique workflows, with a 99%+ activity success rate. More than a third of all merged PRs inside Cursor's own engineering organization now come from cloud agents.
Capital One's Vivek Gupta and Joseph Gilbert demonstrated the modernization of cloud provisioning at scale. Capital One moved from static infrastructure and complex choreographed deployments to a unified Durable Execution model using Temporal Open Source. This was driven by the need to solve troublesome state management, heavy SDLC burden, and decentralized failure handling that had grown unwieldy. Their guiding principles, which were "Code First, Not Config" and "Reliable Rollbacks", offered a practical framework for anyone navigating a similar transformation inside a highly regulated environment.
They closed with a quote from Werner Vogels, Amazon's CTO: "Everything fails, all the time." But as Vivek explained, Durable Execution can help you get ahead of that.
One pattern ran through almost every session's Q&A: rather than asking about technology, engineers focused on organizational change. More specifically, they were curious about how to get teams bought in and how to navigate the internal politics of platform adoption. When I spoke with attendees during the hackathon, in sessions, and beyond, they let me know that Temporal is actually quite easy to understand. The hard part is everything else.
Not just another developer conference#
Replay was an event where you never knew who you might bump into. I met Maxim Fateev, the CTO and co-founder of Temporal, in passing between sessions. A few minutes later, I also introduced myself to Samar. While both conversations were brief, there's something really grounding about meeting the people who built the thing you're betting your career on.
The fun stuff#
Writing about my experience at Replay would be incomplete without describing the Tiki Bar. In a "private" room tucked away from the rest of the event, humans and robots served guests mocktails and boba tea. With a tall stack of vintage computers inside the bar displaying clips, vintage memes, o.g. Internet interfaces (remember Internet Explorer?), and Leonardo DiCaprio as Jay in The Great Gatsby tipping his glass to you... It was such a unique experience.
| Tall stack of computers playing fun content at the Tiki Bar |
I had a mocktail with a member of the Google for Startups team, a sponsor of the event, as he explained the wide range of resources Google provides for startups, including up to $200,000 in Google Cloud credits. We went back and forth discussing career paths that led us here and marveled about the stage of technology we were in the midst of — quite easy to do when a robot is serving everyone drinks!
| Robot making drinks at the Tiki Bar |
And I can't forget the digital badge! I was amazed the moment I saw it. It's custom built with a digital screen that allowed me to play games and hack it to my personal liking, while resembling the kind of personal touch I had with gaming products growing up. Teams across Temporal spent a lot of hard work assembling it, and everyone I talked to about it smiled and thought it was amazing. I'd say it was time well spent.
| Attendee wearing a digital badge |
| Digital badges laid out after being assembled |
Easily one of the most popular attractions on the floor was a giant 8-foot long keyboard at the Braintrust booth. Turns out nothing can prepare you for actually trying to type on one. I failed spectacularly.
| Attendees at the Braintrust booth with the giant keyboard |
The final night closed with a live set from All Time Low, and nobody was winding down. People who had spent three days talking about distributed systems and fault tolerance were dancing. It was a great reminder that conferences, even technical ones, deserve to be fun and focus on the people rather than the product. Attendees were smiling and laughing, as they danced into the night.
| The stage being set before the after party |
| Alex Gargath acknowledges the crowd during the live set |
Wrapping up#
Being at Replay as a new Temporalite was an experience I didn't fully expect to appreciate as much as I did. Some of my favorite moments came from sharing stories with attendees, whether that was from bumping into people in the hallways or watching engineers build at the hackathon. Every time I walked out of a breakout session, I left having learned something that took teams at those companies years to learn and perfect. That in itself was a powerful feeling.
If you're thinking about attending Replay '27, my advice would be simple: go in curious. You don't need to be a Temporal expert. The community is ever-evolving, and the depth of knowledge being shared is rare. I came in nervous and not knowing what to expect, and left having connected with great minds and a clearer picture of where Temporal is headed.
I already feel part of the Temporal community, and I'm really excited for Replay '27!