Nabil Bukhari works as the CTO of Extreme Networks, an American company that specializes in artificial intelligence-based cloud networking. The idea of creating his own AI clone started as a joke, but for almost 8 months now, the digital twin has been performing some of the CTO duties.
Nabil Bukhari works as the CTO of Extreme Networks, an American company that specializes in artificial intelligence-based cloud networking. The idea of creating his own AI clone started as a joke, but for almost 8 months now, the digital twin has been performing some of the CTO duties.
At Extreme Networks, a company with about 3,000 employees, Nabil Bukhari combines several roles at once: he is in charge of artificial intelligence, is responsible for the product portfolio, and holds the position of CTO.
"My team was discussing how we have to sit through all these meetings, and someone said, 'If only we could be in multiple places at once!' We laughed about it. But then I thought, 'Maybe we can actually be in multiple places at once?' And that's when we got serious," Bukhari told Business Insider.
The AI was trained on samples of Buhari's internal and external texts and speeches, including transcripts of meetings, social media posts, public speeches, and press interviews.
“This helps the AI agent not only think like me, but also sound like me. This agent takes all the reports that the teams were going to present to me, analyzes the dialogues the way I would, and asks the same questions that I would ask,” said the CTO of Extreme Networks.
So now, instead of trying to find a free window in the manager's schedule (which can be very difficult), teams work with their AI agent during the first review phase. The agent asks them questions and provides feedback.
"It's sometimes even a little creepy to read, because he writes exactly what I would ask myself," Bukhari noted.
Currently, 80% of project and program reviews don't even reach the CTO.
The team also provides the agent with feedback about the interaction, which is then evaluated to improve the AI.
“At first, it was just reviewing project progress reports, but we expanded its scope. It now reviews software updates, business plans, product specifications, and other similar structured documents. This has reduced the time spent on each project and freed up space in the schedules of all teams.”
At first, the questions asked by Buhari and his AI clone matched by about 50%. Now, that figure is 85–90%.
“However, we still insist on keeping a human in the process. AI is now at a stage where the question is not whether it can make a mistake — it will. Mistakes may be rare, but you never know where exactly they will happen. Therefore, human participation is mandatory, especially when it comes to critical decisions,” Bukhari added.
The team has full control over what advice the agent gives, what decisions they make, and how they justify them. Decisions that exceed a certain threshold of importance are always reviewed personally by Bukhari.
"I am sincerely convinced that this is the model of the future, where there will be no confrontation between 'people and AI' - instead, there will be a synergy between people and artificial intelligence," he believes.
Buhari added that thanks to his clone, he now has more time for his daughters.
