If you’ve been around the DevOps world long enough, you know the DORA report isn’t just another whitepaper. For over a decade, it’s been the yardstick by which teams measure themselves and the north star guiding how leaders think about performance. From the moment DORA introduced the famous four key metrics — deployment frequency, lead time, mean time to restore and change failure rate — it gave us the vocabulary to prove what many of us already believed: that speed and stability aren’t opposites, they’re companions.
But the 2025 edition of the DORA report — focused on AI-assisted software development — hits differently. Not because it rewrites the fundamentals, but because it underscores something the platform engineering community has been saying for years: platforms are the multiplier. Without strong platforms, AI is little more than an expensive parlor trick. With strong platforms, AI becomes a force for real business value.
Platforms: From Backstage to Center Stage
Let’s cut through the noise. A full 90% of organizations surveyed now report having some form of internal platform, and 76% have dedicated platform teams. That’s not hype — that’s critical mass. Platform engineering has gone from a niche experiment to the backbone of modern software delivery.
And DORA 2025 makes the case crystal clear: The quality of your platform directly determines whether AI helps or hurts. When platforms are well-designed, with paved paths, guardrails, and a strong developer experience, AI adoption correlates with significant organizational performance gains. When platforms are half-baked or inconsistent, AI’s impact is negligible.
Truth is, your shiny new AI license won’t save you if your developers are still wrestling with inconsistent pipelines, flaky environments, and ad-hoc tooling. AI amplifies whatever foundation you give it — for better or worse.
Faster But Fragile: The Throughput Trap
The report also highlights one of the most counterintuitive findings: AI improves throughput but still increases instability.
On the surface, that’s cause for celebration. Finally, AI is delivering the speed bump we’ve been promised. But the hidden cost is reliability. If your platform doesn’t provide guardrails — rollback features, test automation, standardized environments — AI-driven speed just means more broken builds, more flaky deployments, and more late-night incident calls.
As I’ve said before, AI is like rocket fuel. Platforms are the guidance system. Without one, you’re just accelerating toward the cliff.
The VSM Connection
Another DORA nugget that resonates with platform engineering: Value Stream Management (VSM) is the force multiplier. AI can crank out local optimizations — faster commits, quicker code generation — but without VSM, those wins never translate to broader outcomes.
Platforms are where VSM becomes real. They’re the mechanism that connects individual productivity boosts to team- and org-level performance. They surface flow metrics, enforce standards, and make sure AI doesn’t just make developers “feel faster,” but actually makes the business move faster.
In other words, VSM without platforms is a spreadsheet. VSM with platforms is a steering wheel.
Trust and the Developer Experience
One of the most telling stats in this year’s report: While 90% of practitioners use AI, nearly a third admit they don’t trust AI-generated code.
That trust gap is a platform problem as much as it is a cultural one. If developers have safety nets — version control, easy rollback, automated testing—they’ll feel more comfortable experimenting with AI. If they don’t, every AI suggestion feels like a potential landmine.
High-quality platforms create the psychological safety engineers need to actually benefit from AI. Without that, you’re just adding stress, not removing it.
User Focus: Platforms as Enablers
Another powerful finding: teams that adopt AI without a strong user-centric focus actually see performance harm.
That one hits home for platform engineers. Platforms aren’t built for their own sake—they’re built to enable product teams to deliver value to end users. When platforms are designed around developer experience and user value, AI becomes a genuine enabler. When they’re designed around internal politics or tool sprawl, AI just accelerates dysfunction.
What it Means for Platform Engineering Leaders
Here’s the bottom line from DORA 2025, through the lens of platform engineering:
- Platforms are the amplifier. Good platforms make AI effective; bad platforms make it irrelevant.
- Guardrails matter. Rollback, version control, automated testing — these aren’t “nice-to-haves.” They’re what let you go fast and safe.
- VSM belongs in your platform. Don’t bolt it on after the fact — build flow metrics and alignment into the platform experience.
- Trust is designed, not assumed. Give engineers the tools and safety nets to trust AI, or adoption will stall.
- Keep the user at the center. Platforms that don’t align with user value just accelerate the wrong work.
A Brief History Lesson
If you think this is all just this year’s AI hype cycle, think back to DORA’s history. In 2014, when the first reports dropped, they demolished the idea that speed and stability were mutually exclusive. They showed with hard data that elite performers moved faster and broke less. That finding alone reshaped boardroom strategies and justified billions in DevOps investment.
Fast-forward to 2025, and DORA is doing it again. It’s telling us that AI doesn’t change the fundamentals. It just makes the fundamentals more important. Platforms, pipelines, feedback loops, user focus — those are still the bedrock. AI just raises the stakes.
Shimmy’s Take
I’ve been around long enough to see this movie a few times. New tech shows up, everyone rushes to adopt it, the hype cycle spins out, and the fundamentals get ignored. Then reality sets in, and the survivors are the ones who had their foundations solid all along.
The 2025 DORA report is a reminder that platform engineering is the foundation. Without it, AI is lipstick on a pig. With it, AI can deliver on its promise — not just faster code, but better outcomes.
So here’s my advice: If you’re a platform leader, don’t get distracted by shiny demos. Invest in your platform quality. Nail developer experience. Bake in VSM. Make trust and rollback easy. Because when the dust settles, that’s what separates the winners from the also-rans.
As DORA has been telling us for a decade, tools don’t fix culture, and speed doesn’t fix bad process. AI won’t save you. But with the right platform, it might just give you the edge you need.