DORA 2025 Thoughts

Hey Folks... a few days ago I posted a some initial thoughts, but this topic warrants more than a quick take. Huge kudos to Nathen Harvey and the entire DORA team The full report is well worth your time: 👉 dora.dev

Every few years a report lands that doesn’t just describe the industry; it names the moment. This year’s DORA report does exactly that. It says, with data and a quiet kind of bravery, what many of us have felt in our bones: AI is a mirror and a multiplier. It amplifies whatever you’ve already built, good or bad. If your delivery system is clean, disciplined, and humane, AI becomes rocket fuel. If it’s brittle and tangled, AI simply accelerates the chaos.

AI is a mirror and a multiplier

That’s the uncomfortable gift of this report. It refuses the fantasy that a tool can redeem a system. It puts the spotlight back where it belongs: on architecture, flow, and the covenant we keep with the engineers who carry our ambitions to production.

Let’s be honest about what changed. The last decade rewarded teams who learned to ship in small batches, automate the boring, and make rollbacks graceful. Now AI has pushed us into a new gear. Code appears faster. Reviews feel lighter. Docs are less of a tax. But the system (the actual system that turns intent into value) hasn’t magically become safer. In fact, AI has revealed exactly where it isn’t.

That’s why DORA lands with a thud for leaders who only count commits. You can’t spreadsheet your way out of coupling. You can’t prompt-engineer your way past an approval gauntlet that takes days. You can’t chat your way around a platform that makes the right thing hard and the wrong thing easy. Speed is a spotlight; it doesn’t fix the set.

Here’s the turn: the report doesn’t scold. It invites. It says, “Design the system.” Treat your platform like a product. Make the golden path the fast path. Minimize the blast radius by default. Bias for reversibility. Collapse the time to signal so feedback arrives while the context is still warm. This isn’t theater. It’s craft.

When you internalize that, AI stops feeling like a gamble and starts behaving like leverage. Teams move from “AI as a nifty sidecar” to “AI as a force multiplier for a well-designed socio-technical system.” That’s the leap. Not more models. More system.

I love the human arc in this, because DORA quietly centers it. We don’t just ship code; we steward attention. We don’t just set targets; we shape the daily experience of building. Platforms, policies, and paved roads aren’t bureaucracy; they are how leaders choose where talent gets spent. Do we spend it on toil, or on breakthroughs? AI raises the ceiling. Our job is to raise the floor.

So what does raising the floor feel like on Monday?

It feels like publishing a one-page AI stance so ambiguity stops taxing every decision. It feels like turning rollback into a ritual: practiced, boring, reliable. It feels like two new golden paths that make “new service in minutes” and “safe deploy in one step” the path of least resistance. It feels like instrumenting the platform with SLOs you’d be proud to show your board: time to first deploy, golden-path adoption, mean time to clarity when something breaks. It feels like mapping the value stream and removing the wait state you’ve tolerated for years because “that’s how we do it here.”

Most of all, it feels like engineering leadership choosing to be designers of experience, not curators of tools.

There’s a deeper pattern emerging, one I’ve been calling the Great Abstraction: the shift from code-as-craft to intent-as-spec and the rise of the Agentic Engineer. We’re moving from typing to orchestrating. From “what’s the implementation?” to “what’s the architecture of possibility?” DORA validates that direction, but adds a sober constraint: intent only becomes value at the speed of your feedback loops and the safety of your reversibility. That’s the physics. Ignore it, and AI just helps you get lost faster.

Some leaders will read the report and buy more assistants. The wise ones will redesign their systems.

They’ll remove coupling they’ve been afraid to touch. They’ll make platform usability a board-level concern. They’ll insist that every heroic workaround become a paved road. They’ll measure outcomes that matter: flow, stability, and user impact. They’ll hold those lines even when the demo looks dazzling. They’ll invest in judgment: how to shape prompts with context, when to trust, when to verify, when to stop.

And something beautiful will happen. The anxiety that comes from “more, faster, louder” will give way to a quiet confidence: we can move quickly and safely because our system was designed to. That confidence compounds. It keeps talent. It invites bolder bets. It turns AI from a risky accelerant into an ambient capability: present, powerful, and almost invisible.

One last thought

In every golden age, there’s a moment where the tools outpace our habits. This is that moment. The report is not a hype train; it’s a mirror. If you don’t like what you see, don’t blame the mirror. Redesign the room.

Build the system worthy of the speed.

Neil Douek - Engineering futurist, platform and DevOps obsessive

Neil is working with a superb team of experts at OTTRA Limited . They helps banks, government agencies, and other regulated organisations modernise their software delivery. We're ISO27001 and Cyber Essentials certified DevOps specialists who transform your people, processes, and technology - turning quarterly releases into daily deployments while strengthening your compliance posture. Visit https://www.ottra.io to find out more.