Mark Cuban is trying, in his words, to "break the healthcare industry" with Cost Plus Drugs, his online pharmacy. He blames a handful of big healthcare conglomerates for much of what is wrong with it, and he now has software keeping watch on them. "Every time they get sued, it comes to me and tells me," Cuban said. The same goes for every contract they bid on.
Cuban, the billionaire investor and former Shark Tank investor, described the setup in an episode of The Diary of a CEO published October 26, 2026. The host, Steven Bartlett, pressed him on what it means for jobs, and Cuban agreed that the work would once have gone to a person.
What the agents do
An AI agent is a program built on a language model that does more than answer a question. It decides which steps to take, uses tools such as search or email, checks the result and moves on to the next step. Anthropic, the company behind the Claude models, describes agents as systems that choose their own sequence of actions and get feedback from their environment as they go. Because one error can carry into the next step, the company recommends checkpoints and human review.
Cuban's monitoring setup is a plain example. One agent alerts him when one of the conglomerates is sued or bids on a contract. He then uses an agent to write an email to request copies of the related contracts under the Open Records Act. This was something he "could never have imagined doing" before unless he hired someone, he said, and hiring meant taking the time to find, train and manage that person, so "it never got done." He did not say which tools run the workflow.
When Bartlett pointed out that Cuban would otherwise have hired people for some of these tasks, Cuban answered: "Yes, that's true." He put a price on it. He might have brought in an intern at "$15 an hour or $20 an hour, whatever it was."
His case is that the trade does not end there. When agents get more done, he said, you have "better opportunities to use what you know to create other things." The time saved might produce a new business idea, and that new business needs people. He said the staff he now hires do work "at a completely different level," and that all of it takes less of his time.
A book in 25 years
Bartlett described his own experience with mixed feelings. He has "I don't know, 25 AI agents on my phone now going through all of my stuff," he said. He called it great, then trailed into an expletive and an "oh my God."
Cuban credited agents with a smaller personal change. He had not read a nonfiction book in 25 years, he said, and the only reason he found the time was that agents had cleared his to-do list: the searching, the reading, "all the tedious bullshit work that you didn't want to do or always couldn't get to or thought you had to hire somebody for."
He named several agent products, including Cowork, Anthropic's agentic workspace for office tasks such as research, reports and spreadsheets across files a user approves. Anthropic's release notes date its first preview to January 12, 2026. Cuban said the agents all give different answers, "and then you have to use your brain again."
Creative destruction in Terre Haute
Bartlett asked about AI executives who say there will hardly be any jobs left. Cuban agreed that AI will transform work but called the result "creative destruction," the old economic idea that new technology wipes out some industries while building others.
His example came from his student years at Indiana University in Bloomington. Nearby Terre Haute, Indiana, he said, was once home to RCA Records manufacturing, which moved on to CDs and then DVDs. That work is "gone, gone," he said, "but it's not like there aren't people making money on music." In his view, new businesses will use the new technology and create different kinds of jobs. He did not promise a smooth transition. "Is the path going to be painful for people? Yeah, just like the people in Terre Haute, Indiana. It's going to be painful."
Tied for second
Cuban's approach to AI goes back to his early days selling software in the 1980s. He reasoned that only the person who wrote a program would understand it better than he did, so if he worked hard enough he was "tied for second." He stayed in the retail store overnight or took the computer home to his apartment and read the manuals over and over, he said. His own blog posts about that period describe the same practice, and how knowing competing packages let him advise customers and pick up consulting work.
He applies the same rule to every new AI model. "There's the creator and everybody else is tied for second," he said, and he intends to win by learning it "faster, better." After a long career in technology, he said, AI "dwarfs everything else in aggregate." He sees change as opportunity, and he expects many people to avoid it, "because you know most people aren't going to do the work."
He traced that attitude to a line he heard in college from Bobby Knight, Indiana University's basketball coach, on 60 Minutes: everyone has the will to win, but only those with the will to prepare actually do. In a 2007 post, Cuban presented his version of the line as a paraphrase from memory.
What he tells his children
Cuban's children are 17, 20 and 23. When his oldest daughter was in college, he told her she had to learn agents. "She was like, yeah, whatever, whatever," he said. Then he watched one of her presentations, in which she described the tasks she had got an agent to do. "I was like, proud papa," he said.
His advice to them is blunt: learn this, because whoever does "is going to be so much more productive than you are." Anyone with a smartphone, or a library laptop with access to Claude or Gemini, has "no question you can't ask," he said. Asked by Bartlett whether hard work still matters, he said yes. People still have to take the time to ask, absorb the answers and learn to ask better questions. Cuban said he never had mentors or a network and learned by reading. Now he skips the book and asks Claude or Cowork to find everything on a subject, or to tell him what Warren Buffett thought about it.
Bartlett suggested that, for ordinary listeners, the safest bet is getting good at something, along with people skills; an agent might run his computer, but he would still have to interview people. Cuban agreed those skills become more valuable. Some of his smaller investments follow that logic, he said, such as a company that runs silent discos, which depend on face-to-face contact. He said the same thinking is part of why sports valuations have risen so much, because people consider sports "AI proof," though he added that it "has other issues."