Transform 2025: Why observability is critical for AI agent ecosystems

The Autonomous Program Revolution is coming. Horse Transform 2025Overcome Willy, New Relict and Sam Witteven, CEO and Red Dragon EU co-founder, How to work for tools for measurable ROI, a tool for tools for tools and infrastructure road map

The new relict provides customers to seize and linking real-time application, access and infrastructure telemetry. Observation goes beyond monitoring – the context of teams is to equip the context of teams and to overcome problems and optimize complex systems, even in front of unexpected issues. Today, this generator and agent are much more complicated as the EU mixture. Now watching the company, Nvidia Nim, Deepseek, Chatrept and all about it reflects the setting of about 30% to use the AI ​​monitoring.

“The other thing we see is a great variety of models,” said Willy. “Enterprises starting from GPT, but have started to use a bunch of models. We saw a 92% increase in the change of models.

Observed in the agent world

In other words, how does the observation develop? This is a big question. The use of use is wildly varies in industrial areas and is radically different for each individual company, depending on functionality, size and goals. A financial firm can be directed to maximize EBITDA margins, a product-oriented company measures the market speed along with quality control.

When a new reliques was established in 2008, the center of gravity for observation was the application of applications for SAA, mobile and then the resulting cloud infrastructure. Agents such as AI and Agentic AI’s rising, agents, micro agents and nano agents, produce and produce AI-written code.

For AI observation

The number of services and the number of microservicities, especially for native organizations, the cognitive burden for native organizations is extremely rising. Of course, AI can help it, Willy says.

“The way you will go to work will be enough information where you will work in cooperative mode,” he said. “Promises of the promise of the observed agents are to take this automatic workload and take place. It will be democratized for more people.”

Observation of a single platform agent

The unified platform for observation benefits from the agent world. Agents automate workflows, but they form deep integrations in all the ecosystems in all numerous vehicles in an organization like GitHub, Servicenow and other. With Agentic AI, the developers can be warned to what happened with code errors anywhere in the codestem and correct immediately without leaving the coding platform.

In other words, if there is a problem with the code placed in GitHub, an observation platform equipped by agents can detect it, determine how it will solve it, or then automates the engineering.

“Our agent looks over the information on our platform,” said Willy. “This can happen to how the application does, the main Azure or AWS structure is something – something related to the placement of this code.

For example, in GitHub, when a developer is working well, when mistakes are managed or the slip of a program, they said it was provided with the further increasing. The next step announced by the new relict last month, together with the Copilot Coding Agent, this issue is clearly explaining this issue. Copilot then returns, corrects the issue and then receives a ready version to reset.

The future of Agentic AI

When organizations began to adopt the Agentic AI and begin to adapt him, it is a functional part of the observation capability, says.

“When you start building all these agent integrates and tracks, you want to know what the agent is doing,” he says. “This is a kind of thinking for the infrastructure. The way of thinking about what is happening in your production. This is what is observed and we are in front.”