Final week, we had our first Infrastructure & Ops superstream of 2026, Platform Engineering within the Age of AI. Our audio system explored a spread of matters centered on supporting new AI workloads, every with distinctive infrastructure wants, unpredictable prices, and novel safety issues. Google Cloud’s Abdel Sghiouar took the viewers by way of what an excellent platform for AI seems to be like, Cockroach Labs’ Jordan Lewis shared classes realized rolling out a company AI platform, Syntasso’s Daniel Bryant outlined a three-layer mannequin for constructing an excellent platform, know-how chief Sarah Wells mentioned the significance of governance and how one can make it extra manageable, and Thoughtworks’ Ben O’Mahony defined why evals ought to be a part of your observability story. You possibly can watch the highlights right here.
The occasion concluded with a fireplace chat between Sam and Nathen Harvey, who leads the DORA crew at Google Cloud. DORA has been monitoring software program supply efficiency for over a decade, which implies they’ve watched plenty of know-how tendencies come by way of. Their heart of gravity has all the time been the identical query: How shortly and safely can a crew transfer change right into a working manufacturing software?
AI hasn’t modified that query, though it has made answering it a bit more durable. DORA just lately launched its ROI of AI-Assisted Software program Growth report to point out how AI is working for groups proper now, and the way which will or might not be contributing to organizations’ backside traces. Nathen used the findings as a jumping-off level to dig into how AI is altering platform engineering and software program improvement as a complete.
The productiveness hole
Sam began by stating one of many greatest headline findings from DORA’S 2025 knowledge: Organizations noticed about 10% enchancment when it comes to precise code shipped to manufacturing techniques. Regardless that builders doubtless felt that they have been extra productive, that doesn’t routinely carry by way of to manufacturing. DORA’s knowledge reveals greater throughput alongside greater instability. In different phrases, groups are delivery extra however they’re additionally extra regularly rolling again modifications or implementing fixes. The positive factors on the particular person degree are actual (and 10% is a reasonably good quantity), however these positive factors aren’t “the dramatic enhancements that you just discover within the headlines.”
AI amplifies good processes (and dangerous ones)
Nathen defined that AI is an amplifier and mirror that equally displays the nice and dangerous. On groups the place delivery change is already straightforward, AI tends to maintain issues working effectively. On groups the place getting turn into manufacturing is painful, AI generates extra change and makes the prevailing friction extra acute. That mentioned, his learn on this end result is cautiously optimistic: “If the ache is extra acute, we possibly will put money into addressing that ache.”
The rub is that the funding has to truly occur. Nathen famous that in lower-performing organizations, AI instruments usually arrive with a reset of expectations slightly than an invite to repair the method: Right here’s your new software. Now we anticipate extra from you. Addressing this downside means reframing the query “Does AI make folks extra productive?” What we actually ought to be asking is “Underneath what situations will AI enhance productiveness, and who’s liable for creating them?” And that falls on the group, not the know-how.
Verification isn’t a checkbox
Belief is a giant problem with generative AI. About 30% of DORA survey respondents belief AI output little or in no way. Round 46% belief it “considerably” (and Nathen is considered one of them). Regardless of all of the advances in generative AI, these instruments nonetheless make errors, and when you’ve multiplied your potential to generate code with out doing something to scale your potential to confirm it, you’ve made your scenario worse, not higher.
Nathen known as this the verification tax, and it belongs in any trustworthy accounting of AI’s productiveness impression. Pipeline adaptation belongs there too: Is your supply pipeline match for goal given the quantity of change you’re now making an attempt to push by way of? These prices don’t present up within the headlines about 10x developer productiveness. They present up in your incident studies three months later.
DORA just lately printed an ROI framework and calculator for AI-assisted software program improvement. Nathen was clear that there’s no common quantity to supply, and the calculator doesn’t fake in any other case. What it does is give groups a method to mannequin the true prices, together with the training funding, the verification overhead, and the pipeline modifications required.
Context switching and burnout
With productiveness on the upswing, AI-induced burnout is changing into a critical concern. (Steve Yegge calls this the “AI vampire.”) DORA’s knowledge for 2025 confirmed that AI adoption wasn’t strongly related with burnout, with the caveat that about 64% of DORA survey respondents mentioned they’d by no means labored in an agentic workflow. Each of these findings are more likely to change considerably in 2026.
Nathen highlighted one supply of burnout he expects to escalate as brokers turn into the norm: context switching. As he identified, software program builders spent years arguing for protected focus time to do the deep work that requires them to keep up circulate. Agentic workflows are actually incentivizing those self same builders to voluntarily run a dozen or extra brokers without delay, forcing them to context-switch a number of instances each hour. As he joked, “There’s loads of analysis that helps the concept all of us really feel like we’re fairly good multitaskers and none of us are.” The results are coming, and we’re doing it to ourselves.
The cognitive debt query
Sam Newman introduced up the associated notion of “cognitive debt,” and specifically, Margaret-Anne Storey’s dialogue of it. (See “How Generative and Agentic AI Shift Concern from Technical Debt to Cognitive Debt” and “From Technical Debt to Cognitive and Intent Debt: Rethinking Software program Well being within the Age of AI.”) Right here’s how Storey explains the issue in her weblog put up:
Debt compounded from going quick lives within the brains of the builders and impacts their lived experiences and skills to “go quick” or to make modifications. Even when AI brokers produce code that may very well be straightforward to know, the people concerned could have merely misplaced the plot and should not perceive what this system is meant to do, how their intentions have been applied, or how one can presumably change it.
And as Sam famous, this compounds throughout groups and organizations. As builders more and more work in parallel with AI slightly than with one another, they lose the shared understanding that comes from folks constructing software program collectively. Kent Beck as soon as mentioned that “software program design is an train in human relationships.” Agentic workflows are placing stress on that in methods we’re solely starting to see.
Nathen agreed cognitive debt is the place he’s most involved, and each your employees and your structure will endure for it. Understanding the ramifications of an architectural determination you made eight months in the past takes years of operation to floor, and AI doesn’t assist with that in any respect.
Put money into your platform now
Contemplating what makes some AI-assisted groups excessive performers, Nathen defined, “It’s not that you’re utilizing AI however how you’re utilizing AI.” This commentary led DORA to develop seven capabilities that, when mixed with AI adoption, result in higher outcomes. Nathen briefly ran by way of the checklist, ending on high quality inside platforms. And right here he made a declare about software program engineering funding that was, in his phrases, “a bit bit wild”:
Each product engineer that you’ve in your group, each engineer that’s centered on constructing options proper now, ought to most likely cease constructing options and deal with the platform.
His argument is that platforms matter extra, not much less, in an setting the place AI makes it attainable for nearly anybody in a corporation to construct one thing. The folks closest to clients and enterprise issues can now generate working software program. What they will’t do is be sure that software program is sturdy, safe, and production-ready.
Nathen prompt that the most effective leverage for software program engineering funding at present is likely to be constructing platforms that present these guardrails, that shift the complexity of production-readiness down into the infrastructure in order that anybody constructing on prime of it will get the protection web without spending a dime. He acknowledged that shifting each product engineer to platform work is likely to be overkill. However the path of journey is actual. The platform can also be, as Newman identified, the place you convey determinism again right into a course of that AI has made extra nondeterministic.
That’s one thing we’ve been listening to so much right here at O’Reilly. The growth of who can construct doesn’t cut back the necessity for deep engineering experience. It modifications the place that experience is most precious, and platforms are an excellent reply to the place.
What DORA’s analysis tells us
The groups which might be doing effectively are working experiments, studying from them, and spreading these classes. The measure Nathen prompt just isn’t what number of tokens you’ve consumed however what number of experiments you’ve run and the way effectively you’re distributing what you’ve realized.
The instruments are shifting quick sufficient that any group locking in a hard and fast coverage round particular instruments will discover itself caught. What you need is the capability to continue to learn, which implies constructing the tradition and the processes that make studying seen and transferable.
All of DORA’s analysis is freely accessible at dora.dev, together with the 2025 annual report and the ROI framework. The DORA Neighborhood gives an area for practitioners to work by way of these questions collectively. Should you’re making an attempt to navigate any of this along with your crew, you could wish to spend a while there.
And if you wish to dive deeper into Nathen and Sam’s chat or discover the opposite classes, you’ll be able to watch the complete Infrastructure & Ops Superstream on the O’Reilly studying platform. Our subsequent occasion, on September 9, will cowl agentic observability. Register without spending a dime right here, and take a look at all the opposite free stay occasions on O’Reilly.
