Technical Context
I took a close look at the DORA 2025 findings and was struck not by the usual AI hype, but by a phrase that hits the nail on the head: AI acts as an amplifier. If a team is already well-organized, processes are stable, and the platform is robust, AI adds speed and value. If it's chaotic internally, that chaos just starts moving faster.
The report moves away from the old flat scale of 'elite' or 'low performer' and describes seven team profiles. In practice, the most interesting contrast is between Harmonious High-Achievers and teams like Legacy Bottleneck or Survival Mode. For the former, AI boosts throughput, efficiency, and value delivery. For the latter, it increases local productivity, but these gains die in testing, releases, security, and manual approvals.
Honestly, this is very similar to what I see in real projects. A developer gets copilots, LLMs, and code generation, but then the code hits a wall with an old CI/CD pipeline, a weak platform team, a lack of proper guardrails, and endless delays between departments. In the end, it seems like AI was implemented, but there's almost no business result.
The numbers also tell a story. DORA reports that about 90% of organizations already have platform engineering capabilities, and 76% have dedicated platform teams. But having a platform isn't a silver bullet. Its quality is what matters: can it actually translate individual AI wins into systemic results, rather than just adding another layer of bureaucracy on top of Kubernetes?
Another important point: AI most significantly boosts individual efficiency. That's logical. A person writes, searches, and drafts solutions faster. But friction and burnout don't just disappear. Sometimes, it gets even more 'fun' in a bad way: the team produces more changes, and system instability grows.
What This Means for Business and Automation
For me, the main takeaway is very down-to-earth: mass AI adoption cannot be planned like a software license purchase. It's an architectural and organizational effort. If a team is constantly putting out fires, a new AI tool will most likely just add more fuel.
The winners are those who already have solid platforms, clear lines of responsibility, measurable delivery metrics, and sane change processes. Such teams can quickly turn AI into real AI solutions for the business: accelerating development, support, analytics, internal knowledge systems, and agent-based scenarios. For them, AI doesn't get stuck halfway between a demo and production.
The losers are companies that try to build AI automation on top of a fragile operational foundation. I've seen it many times: they want to create an AI agent for support or sales, but internally there's no clean data, no SLAs, no process owner, not even a unified logic for handling requests. An agent in such an environment is not a magician. It quickly exposes all the holes that were previously hidden by manual labor.
Therefore, I would read DORA 2025 not as a report about tools, but as a report about system maturity. In short: first, get the team out of Survival Mode, then scale AI. Sometimes the best first step in AI architecture isn't a new LLM, but rebuilding a workflow, establishing proper observability, creating a platform for safe experiments, and drastically simplifying manual handoffs between functions.
This is exactly what we at Nahornyi AI Lab work on in practice. We don't just bolt a model onto an interface; we look at where AI will truly provide leverage and where you first need to integrate artificial intelligence with data, processes, and the platform, without any illusions. Otherwise, you can automate chaos perfectly. That happens too. It's just not very useful.
This analysis was written by me, Vadim Nahornyi of Nahornyi AI Lab. I specialize in AI automation, custom agents, and the architecture of AI solutions in real-world business processes, where the working result matters more than the wow effect. If you want to discuss your case, order AI automation, create an AI agent, or build an n8n automation for a specific task, contact me, and we'll break down your project in a human-friendly way.