Artificial Intelligence has officially crossed the hype threshold. 2023 was experimentation. 2024 was excitement, fear, debate, chaos. But 2025? It’s the year reality kicks in.

Today, adopting AI is no longer a strategic debate — it’s a baseline expectation. And yet, everyone seems to be asking the same question:

That’s why this year’s DORA State of AI-Assisted Software Development Report is such a big deal. With insights from nearly 5,000 technology professionals and over 100 hours of qualitative interviews , it cuts through assumptions and reveals what’s really happening inside engineering organizations embracing AI.

And the most important insight?

This is the DORA 2025 headline. It’s simple. It’s uncomfortable. It’s true because they have all the data to back this up.

Think of it like giving someone a sports car. If they already know how to drive well, they’ll go faster, safer, and farther. But if they don’t… well, they’re likely to crash. And fast. AI does the same to your engineering organization.

1. If your workflows are clean → AI accelerates you.
2. If your architecture is modular → AI becomes a force multiplier.
3. If your processes are chaotic → AI generates more chaos, just faster.

DORA calls this phenomenon “localized pockets of productivity lost to downstream chaos.” Which… is painfully accurate. 

According to the report:

1. 90% of developers now use AI as part of their work.
2. 80%+ believe it improves productivity.
3. Yet 30% say they don’t trust AI-generated code.
   

And honestly? That’s healthy. This “trust but verify” mindset shows that engineering teams are maturing in their AI usage. We’re no longer star-struck by perfectly formatted answers. We’re learning to guide AI, validate output, and think critically. AI doesn’t make bad developers good. But it makes good developers even better — if they know how to use it.

This is my favorite plot twist in the report.

We’ve all experienced that moment when AI helps you ship faster — and then… everything breaks. DORA confirms this at scale.

From the above image AI adoption:

1. Increases throughput (good!)
2. Decreases stability (not good!)

Why? Because teams are pushing out more changes, faster. But their systems — architecture, testing pipelines, feedback loops — haven’t caught up. It’s like upgrading the engine without upgrading the brakes.

The result? More deployments, more rework, more production failures.

This is the exact “AI acceleration problem” Gene Kim refers to: unless your control systems (people + processes) are fast enough, AI simply pushes your team beyond its stability threshold. 

One of the most insightful additions this year is the seven cluster profiles of engineering teams. This goes beyond metrics and reveals the human reality behind performance.

Here are the archetypes

Burnout high. Friction high. Everything feels like survival mode. If AI is introduced here, it amplifies chaos.

Systems are unstable. Teams spend all their time firefighting.

Stable systems, but slow, bureaucratic workflows. Burnout from meetings, not bugs.

High talent. High effectiveness. But extremely slow deployments → fragile systems.

Steady teams who move slower but with discipline.

Balanced teams that deliver consistently without drama.

The unicorn teams — high throughput, low burnout, high collaboration, strong product outcomes.

These clusters help you diagnose your real problem:

1. Is AI failing because of technical debt?
2. Team burnout?
3. Process bottlenecks?
4. Architecture issues?

Without knowing your archetype, AI adoption becomes a guessing game and a hard wall to climb.

One of the most striking stats:

That means internal platforms are no longer optional — they are the foundation for enabling AI at scale.

A good platform:

1. Reduces cognitive load
2. Standardizes workflows
3. Gives teams fast feedback loops
4. Enables safer deployments
5. Accelerates AI tools’ effectiveness

A bad platform?

It just becomes a bottleneck that AI only magnifies.

Teams with strong internal platforms unlocked the highest returns from AI. Those without them… didn’t.

This year introduces something every engineering leader should tape to their wall: The DORA AI Capabilities Model.

It outlines seven foundational capabilities required for successful AI adoption, including:

1. A clear AI policy
2. A healthy data ecosystem
3. A strong internal platform
4. A user-centric focus
5. High engineering discipline
6. Quality architecture
7. Fast feedback loops

Notice what’s missing? “Buy a better AI tool.” Because tools don’t fix broken systems. They only expose the cracks.

DORA emphasizes that VSM (Value Stream Management) is the missing bridge between: AI-driven local productivity → actual organizational performance

Without VSM, AI can make individuals faster, but the system still slows them down.

With VSM:

1. Bottlenecks become visible
2. Workflows become predictable
3. Flow efficiency improves
4. AI output translates to real customer value

In their words, VSM becomes a “force multiplier” for AI investments.

The most philosophical part of the report is “The AI Mirror” chapter. AI reveals who you really are as an engineering organization.

1. If your data is clean → AI thrives.
2. If your architecture is modular → AI fits naturally.
3. If your processes are modern → AI accelerates them.
4. If your culture supports learning → AI becomes a growth engine.

But if not… AI doesn’t solve your problems. It amplifies them. This is not a bad thing. It’s an invitation to improve!!

After reading the 2025 DORA Report, one idea kept echoing in my mind:

If your team prioritizes:

1. clean architecture
2. fast feedback loops
3. platform quality
4. healthy engineering culture

AI will feel like a superpower.

But if not, the problems don’t disappear — they simply appear faster. AI does not replace engineers. But AI will replace engineers who refuse to adapt. This might sound cheesy and you might feel like you are hearing this more often, but with this new AI era, it fundamentally changes how we operate, similar to what the food ordering apps did to the restaurant business (but on a much much larger scale)

AI is here. AI is powerful. AI is transformative. But only if we are ready for it.

So here’s my question to you: Where does your team fall in the seven archetypes? And what would AI amplify today if you introduced it tomorrow?

If you have thoughts — or if your team is experiencing this shift firsthand — feel free to write it in the comments I love hearing how others are navigating this new AI-accelerated world.