MSP Documentation Is Killing Margins And Holding Back AI: Lexful CEO

Bolting a ‘general LLM into existing documentation will not reliably work,’ says Lexful CEO Pinar Ormeci. ‘You have to start bottom-up.’

For too many MSPs, poor documentation practices are hurting profitability while also preventing the capturing of massive opportunities in AI and agentic, according to Lexful CEO Pinar Ormeci.

“You run the world’s IT, but on documents that are out of date, scattered, or that live in people’s heads as tribal knowledge,” Ormeci said during a session Tuesday at XChange August 2026. “You think that is a technology problem—but actually, your documentation [practices] are killing your margin.”

[Related: Biggest MSP Challenge Isn’t Security—It’s Capacity: Cavelo Exec]

Too frequently, MSPs are seeking to rely on data that is scattered across multiple systems, she said.

At the same time, much institutional knowledge is only stored in the heads of experienced employees, Ormeci said, which can be a massive issue when a senior staffer leaves the company or takes time off.

These challenges have become a bigger issue amid the shift to deploying AI—which needs data that is trusted and structured, she said.

“AI has fundamentally changed the role of documentation,” Ormeci said. “Now it is the knowledge layer that powers your AI. It is the context behind your automation and agents. And it is the foundation for faster decision-making and execution at scale.”

Without a doubt, Ormeci’s assessment reflects an all-too-familiar issue faced by many MSPs, according to Larry Hall, president and CEO of Dallas-based Qnet Managed Services.

“For most of us MSPs, documentation is scattered all over the place,” Hall told CRN. “When you get new techs in and techs leave, they leave with [the documentation] somewhere that no one knew about. So it’s a lot of missing opportunities there.”

On the other hand, providing technicians with access to an AI agent that utilizes accurate documentation could make the MSP far more productive, he said.

“If there was a way that they could just ask the AI agent—which keeps track of that knowledge—and it could just tell them [the answer], now you’re more efficient,” Hall said.

Ultimately, the shift MSPs must make as they prepare for AI is that documentation must be shifted into an actively maintained system, with capabilities for serving both human workers and AI, according to Ormeci.

And crucially, bolting a “general LLM into existing documentation will not reliably work,” she said. “You have to start bottom-up.”