SRECon Americas 2025

An impressionistic take on SRECon Americas '25

Niall Murphy

April 1, 2025

Those of us who do booth duty at a conference don’t often get a chance to attend many, if any, sessions. So my view on SRECon is necessarily limited by the fact that I spent more time talking to prospects, customers, and people generally in the hallway track than attending the other tracks. Having said that, I still have some impressions I’d like to share.

AI and SRE. AI was everywhere. In the plenaries, the discussion tracks, the BoFs, the exhibitor hall - everywhere. But contrasted with previous conferences, where I felt that in general the small-c conservative nature of production engineering was keeping anything AI-ish at arm’s length, this conference I felt the audience was actually engaging with it.

Or to put it another way, we’ve moved from denial to bargaining about AI.

To be clear, this engagement isn’t without trepidation. Job losses are still happening; most folks in the profession struggle with their organisation understanding the value of the role and their own personal contribution. It’s been said the numbers around investment in AI only really make sense if there’s widespread displacement of human effort. Those of us who spend most of our time in a cost center rather than a revenue center have made our peace with this, but the overall direction of travel is no less stark for being reconciled to it. Many of the conversations I were lucky enough to hear were also asking fairly obvious questions about what happens if (say) most incidents are taken care of by a machine, then how do humans become educated about how the system works overall? To which there wasn’t much of a considered answer.

As distinct from the hallway or the discussion track, the formal talk tracks tended to focus on the mechanics of making this stuff work, rather than actively envisioning a positive future. I will say that I did encounter people who hoped more sophisticated automation would take care of certain kinds of production toil, and maybe even help with some of the more panicked moments of the on-call experience, but I didn’t see much in the way of cheerleading here. Neither, in fairness, did I expect to.

It was a little different in the exhibitor hall, where there was a noticeable uptick in commercial activity and energy, which we noticed both at our booth and also with the presence of Meta and Apple (both there for recruiting). There was a definite buzz around the AI-for-SRE booths, though as a business leader in an adjacent market interested in relevant developments, it was hard from the outside to see much in the way of differentiation. If technical differentiation proves hard, I would expect to see people do more marketing, push customer service, etc, but I didn’t see any of that - yet.

Resilience in software. Perhaps unsurprisingly, given Dr. Laura Maguire’s involvement as co-chair, the Resilience in Software folks were out in force. As AI increases in importance, the movement that seeks to understand the holistic nature of sociotechnical systems, and the adaptability of humans as a positive force is also rising in prominence. This made its way onto the schedule in a bunch of places, particularly Dr. David Wood’s talk, but to my mind was most notable in the discussion tracks and the BoFs. (I co-led one on AI & SRE with Lorin Hochstein, but Rachel Silber, John Allspaw, and many others were also leading and contributing to discussions.) Colette and others were doing a great job guerilla marketing by handing out badge stickers.

My personal opinion is that there are a lot of very useful ideas in this space that the median of SREs are still broadly unfamiliar with, and that could be profitably introduced to ICs - but there are still many struggles with explaining the precise benefits of these approaches to engineering leadership, and that will remain a limiting factor for adoption until such time as a reasonable framework is found.

Overall takeaways. It was a well-organised conference, with a great programme that touched on the key issues, and some stellar speakers and delegates. There was also a nice energy given to it by the high proportion of new attendees. But I can’t help but think it was in some sense a wait-and-see conference, more concerned with what’s going to happen to us in the future than helping to determining it. I trust the between-conference time will advance this discussion and we’ll have more to say in the next one.

Niall Murphy

Niall Murphy is an award-winning, best-selling author, speaker, industry leader, and executive. He has been working in Internet infrastructure since the mid-90s, in a variety of roles from systems, network, and software engineering, and ranging across individual contributor to director scope.

He is best known for the Site Reliability Engineering book and associated works.

He lives in Dublin, Ireland with his wife and two children, and holds degrees in Computer Science, Mathematics, and Poetry Studies.

share