
At OpenText, the constraint is no longer automation replacing roles, but a shortage of people who can take on AI-related work fast enough. As AI adoption accelerates across engineering, QA, and product functions, the company is reallocating talent internally to meet demand rather than reducing headcount.
“We’re not looking for AI agents to replace humans. We want AI agents to augment our human colleagues and allow them to execute better work that is more creative,” noted Muhi Majzoub, Executive Vice President and Chief Product Officer of OpenText.
Greater efficiency
Majzoub highlighted their quality assurance (QA) department as a clear example of AI’s operational impact. While automation already supports part of the workflows, the introduction of AI is expected to double efficiency.
“If I have a team of 30 QA engineers, I may be able to move five or eight of them to become automation experts instead of manually clicking on screens and following a paper script in front of them or a Word document on their next monitor,” he explained.
In addition to QA, the software company has also deployed its Aviator agentic AI platform across technical writing and CI/CD deployment. The Aviators, Majzoub said, run and attach themselves to different platforms, content, items, and business networks for supply chain, customer experience, and cybersecurity.
“Instead of the engineer looking at a 50-page report, they look at an executive summary that may be just a page in length, allowing the engineer to be more efficient and do more work,” he noted.
Human in the loop
To ensure that the AI agents have the necessary guardrails and will solve what customers want to be fixed, the company conducts random testing because as much as AI is wonderful, it is far from perfect, Majzoub said.
“When we rolled out the latest release of our identity management, it ran for four-and-a-half months internally at OpenText, supporting 22,000 employees and hundreds of devices with fingerprints, retinal or facial recognition, two-factor authentication, and all of the above. We ran it, fixed the few issues that we couldn’t catch in AI before we made it available to our customers. Now we have hundreds of customers running that release, including companies like BMW that use our identity management platform to protect three quarters of a million IDs for their dealers and their full-time employees or contractors,” he revealed.
Majzoub also explained how they operate their feedback loop across the industries they serve: “If a customer in the pharmaceutical industries in India or in the United States is demanding something, then I have experts who are pharmaceutical-trained product managers, who are outbound-facing business experts. They’re not coders, they’re not writing code, but they could take a customer requirement and translate it to a roadmap item that comes to engineering for execution.”
Building trust
Down the line, Majzoub expects AI to continue advancing across multiple fields, pointing to generative AI model companies releasing new versions every three to four months. This, he said, will allow enterprises to address more complex business use cases.
Presently, OpenText has between 7,000 to 9,000 engineers globally, and yet there is still not enough manpower to accomplish everything in their project backlog, he noted.
“The advancement in AI is wonderful because as we get more efficient, I’m going to find 10 engineers to move to a new product, another 20 to go build and research a completely new area. For me, that’s an opportunity. Today, we have dozens of agents. I guarantee you, come June or July, we will have hundreds of agents,” Majzoub said.
Meanwhile, he also addressed concerns that AI could replace software development jobs, given that it is now capable of writing code.
“We have a 30-year history supporting some of the public sector agencies in the top 20 western world governments, from India to Singapore to Australia, New Zealand, Japan to Canada, US, Brazil, Mexico, and the Americas, to every one of the countries in Western Europe. In every industry, AI will never replace that trust, or at least it will never replace that for as long as I live,” Majzoub added.













