Sovereign software revolution is coming. at all Change 2025Ashan Wali, CEO of the Reddragon AI, CEO and co -founder, Ashhan Wali, talked about how he was developing an agent system for measuring ROI and charting the infrastructure roadmap to maximize agent AI.
The new Oceshis provides users in real time by obtaining and connecting and connecting to the application, log, and infrastructure. Observations are beyond the supervision – this is about understanding complex systems, solving problems and improving teams, despite unexpected issues. Today, it has become a very complex move that Generative and Agent AI is in this compound. And observed for the company now include monitoring NVIDIA NIM, Depsteek, Chatgpt and everything – its AI monitoring is about 30 % compared to the quarter, the quarter in the quarter, which reflects the acceleration of adoption.
“The other thing we are seeing is a huge diversity in models,” said Wali. “Enterprises started with GPT, but are starting to use a whole group of models. We have seen a 92 % increase in the changes in the models used. And we are starting to see businesses adopting more models. The question is, how do you measure the effectiveness? “
To observe an agent in the world
In other words, how is the observation getting ready? This is a big question. Use matters are brutally different in industries, and in terms of size and goals, the functionality for each individual company is basically different. A financial firm can be focused on the maximum Abbotta margin, while a product -based company is measuring speed in the market, along with quality control.
When the new relief was founded in 2008, the center of gravity for observation was to monitor the application for sauce, mobile, and then cloud infrastructure. The rise of AI and Agent AI is bringing the observation eligibility back to requests, as agents, micro agents, and nano agents are running and developing AI-written code.
AI for observation
As such, the number of services and microsaries increases, especially for digital local organizations, the academic burden for observation of any human handling is greatly increased. Of course, AI can help him, says Wali.
He explained, “The way it is going to work is going to get enough information to you where you will work in operative mode.” “The promise of agents in the observation is that some of them should automatically carry the burden of work and carry out the burden of them. Which will make it democratic to more and more people.”
Single platform agent observed
A single platform agent to observe the world benefits from the world. Agents automatically make workflows, but they make deep integration throughout the ecosystem, in which an organization is in multiple tools, such as harms, gut hubs, services, and so on. With Agent AI, developers can find out what is happening with code errors anywhere in the ecosystem and without leaving their coding platform, they can be fixed immediately.
In other words, if there is a problem with the code in the gut hub, an observed platform run by agents can detect it, set a way to resolve it, and then inform the engineer – or the process is completely automatically.
“Our agent is basically watching every piece of information on our platform,” said Wali. “Whatever the application performance, how the basic Azure or AWS structure is performing, whatever we think this code is related to the deployment of this code. We call it an agent skill. We do not rely on any third party to know APIS.”
For example, in the gut hub, he told a developer when the code is going well, where the errors are handling, or even when the software rollback is necessary, and then automatically make this rollback with the developer approval. The next step, which was announced last month, is working with the Coopelot coding agent to tell the developer exactly what lines of the code are looking at the issue. The co -pilot then goes back, fixes the problem, and then gets the ready version to be deployed.
The future of agent AI
Wali says that since organizations adopt agent AI and start adapting to it, they will know that observation is an important part of its functioning.
He says, “When you start to make the integration and pieces of all these agents, you want to know what the agent does.” “This is a kind of argument for infrastructure. It is argued to know what is happening in your production. The same thing will be seen, and we are at the forefront of it.”