Autonomous software revolution is coming. On Transform 2025Ashan Willy, General Director of Novel Relic and Sam Witteveen, CEO and co -founder of Red Dragon Ai, told about how they base agency systems to a measurable roi and map the infrastructure map to maximize the agency AI.
Novel Relic provides observation to customers by capturing and correlating the telemetry of the application, journal and infrastructure in real time. Obserability goes beyond monitoring – it is about equipping teams in the context and insight needed to understand, solve problems and optimize convoluted systems, even in the face of unexpected problems. Today it has become a much more convoluted undertaking when the AI generative and agency are combined. And the company’s observation now includes monitoring everything from NVIDIA it, Deepseek, ChatgPT, etc. – the employ of its AI monitoring increased by about 30%, a quarter above the quarter, reflecting the acceleration of adoption.
“Another thing we see is a huge variety in the models,” said Willy. “Enterprises started with a GPT, but they start using a whole group of models. We saw a 92% increase in variance of models that are used. And we start seeing how companies accept more models. The question is: how to measure effectiveness?”
Observable in the agency
In other words, how does obserability evolve? This is a huge question. Cases of employ vary significantly depending on industries, and functionality is fundamentally different for each company, depending on the size and goals. A financial company can focus on maximizing EBITDA margins, while the product -oriented company measures the speed on the market next to quality control.
When a up-to-date relic was founded in 2008, the means of gravity of observation was monitoring the SaaS, Mobile applications, and then the cloud infrastructure. The creation of artificial intelligence and agentic AI restores observability for applications, because agents, micro-agents and nano-agents act and produce the code written by AI.
And for observation
With the enhance in the number of services and micros services, especially in the case of digitally native organization, the cognitive load for all tasks of human service service becomes overwhelming. Of course, AI can assist in this, says Willy.
“The way it works is that you will have enough information in which you will work in cooperation mode,” he explained. “The promise of agents in observation is to take some of these automatic loads and make them happen. This democratizes more people.”
One aggressive platform observation
A single observation platform uses the agency world. Agents automate work flows, but create deep integrations with the entire ecosystem, in all many tools that the organization has, such as harnesses, github, service and so on. Thanks to Agentic AI, programmers can be notified of what happens with code errors anywhere in the ecosystem and repair them immediately, without leaving the coding platform.
In other words, if there is a problem with the code implemented in GitHub, the observation platform powered by agents can detect it, determine how to solve it, and then warn the engineer – or completely automate the process.
“Our agent basically looks at all information on our platform,” said Willy. “It can be everything from the application of the application, how the basic structure of Azure or AWS – everything we think is important for the implementation of the code. We call it agency skills. We do not rely on the outside side that learned the API interfaces and so on.”
For example, in Github they inform the programmer when the code works well, where the errors are celebrated, and even when the software is necessary to withdraw, and then automate this withdrawal, with the approval of the programmer. The next step that has announced Novel Relic last month cooperates with the Copilot coding agent to tell the programmer exactly which line of the code is a problem. Then Copilot returns, corrects the problem, and then prepares the version for re -implementation.
The future of the AI agency
Willy says that because organizations accept agent artificial intelligence and begin to adapt to it, they will find that observation is a key part of its functionality.
“When you start building all these agency integrations and elements, you’ll want to know what the agent is doing,” he says. “This is the reasoning for infrastructure. Reasoning to find out what is happening in your production. This will bring observability and we are in the foreground.”