I am writing this on the train home from Budapest where I spent the last 2 days observing and discussing big changes that are currently happening.
On the one side, a change that is coming to Hungarian citizens after their recent governmental elections. Every conversation I had eventually covered the latest political news that was discussed very passionately and with lots of excitement.
On the other side there is the change that has a global impact for all of us working in software engineering. We are going through a major transformation from a Human-Led Software Delivery Lifecycle (SDLC) towards an Agent-Led AI Delivery Cycle (AIDLC) where the role of the human is shifting towards specifications and supervision while the Agents are focusing on building, testing, deploying and operating software as you can see from the following illustration:
The evolution of software delivery towards an AI-First model with different responsibilities between humans and agents
Observability Days Hungary 2026, organized by Provice, was the place where local and international attendees and speakers connected to discuss this change, or revolution, that is happening so rapidly for many of us.
I was honored to be invited back to Budapest to open with my Keynote “The Role of Observability as Cloud-Native becomes AI-Native - and as Business adopts AI-First”:
Opening Observability Day Hungary 2026 talking about the change towards becoming an AI-First company
The key points of my presentation were:
- We are moving with different speeds towards becoming AI-First organizations
- Like with Cloud Native, we will see a bi-modal approach to adopt AI-Native
- Observability is needed for both humans and agents to make better decisions
- Dynatrace supports the journey towards AI-First with several AI-Native Use Cases
From Developer Efficiency to AI Efficiency
I started my talk though with a flashback: In 2025 I presented on Platform Engineering and how Platform Teams can measure how they improve Developer Efficiency. We talked about metrics such as Reduction of Manual Tasks, Time of Developers spent in Productive Tasks or Developer Efficiency.
This year – 2026 - I discussed some of the new KPIs that organizations must use to measure their successful adoption of AI and Agentic Systems:
- Agent Efficiency Index: How many steps does it take an Agent to get a certain type of job done and how does this change over time?
- AI Economics: What is the cost vs output of those agents?
- Human Override Rate: How often does a human need to correct the suggestion or the work of an AI?
It was interesting to see the reaction of people in the audience. Some of them are already on their AI-Native journey using coding agents or agents to support IT Operations optimizing incident response or automating release validation. Some of them – constraint by existing regulations – still waiting for finally making their first steps. Both of those groups understood the importance of those KPIs and the role that Observability plays in getting those insights but also making humans and agents more effective.
Observability Use Cases on the AI-Native Journey
I concluded my presentation with 6 AI-First Observability-Driven Use Cases for different roles across the (S/AI) Delivery Lifecycle that every Dynatrace SaaS user can use today.
Each use case increases elevates observability data toward delivering insights to make better and faster decisions
For each of the use cases I demoed how observability data (logs, metrics, traces …) can be turned into insights (e.g: understanding the business impact of a problem and how to fix the technical root cause). I also highlighted that observability insight consumption moves away from classical dashboards towards being consumed by humans in their respective tools such as their IDE, Git or Ticketing System as well as being consumed by Agents through MCPs or CLIs such as dtctl (Dynatrace’s CLI for Humans and Agents)
If you are interested in learning more about those use cases make sure to follow me on LinkedIn where I have already started to post videos of those demos such as this one for Optimizing Cloud Spend.
A Panel Discussion on Observability
I was also honored to be part of a panel discussion that was led by Milena Kolodziej where we discussed the following questions. I thought to post them here for everyone so that you can think of how you would have answered:
- What is the biggest misconception regarding observability?
- How do you see AI changing the way we use observability?
- At what point does observability stop being about data and start influencing real decisions - performance, cost, reliability?
- Can you share a memorable story where observability met real life - a moment where it made a real difference?
To hear our answers and discussion please keep an eye out for the recordings that I am sure will be posted soon.
Wrap up: a special moment!
I want to conclude my recap with a moment that was very special to me. Before my keynote I was approached by Tibor Jakab and his wife. I met him last year when he attended Observability Day Hungary. This year he decided to make a big step for him: The step on a stage to present his own use case of using Observability at MAV – The Hungarian State Railway. He thanked me for being an inspiration and I was then able to return the favor by giving him a signed copy of my latest book “Observability in the AI-Native Era”.
Great to know the next generation of speakers is on the way. Congrats Tibor for your first talk. Hope you enjoy the book
I know how hard it is to step on a stage for the first time. It’s also hard the many times afterwards. I was just very happy to hear that he made the toughest step – the first one – on this journey. I am sure he will become the one that will then inspire others to also step outside their comfort zone.
Hungary, it was a pleasure! Cu next time!
