Duncan Davidson, an independent consultant and longtime software engineer, joined Ben Lorica on the Generative AI in the Real World podcast to discuss personal AI memory. According to the episode summary, AI tools now remember their users, but each tool keeps a different version of the same person.

Davidson described his own workaround. He first had his coding agents save useful information into Markdown files. After commercial agents added memory, he found that "Claude knew me one way and ChatGPT knows me another way." He could push context back to each tool with instructions and reminders, but he said that takes more tinkering than most people want to do. He wants a small set of shared mechanisms that let a person carry context across services.

He also said that portable memory needs boundaries. He keeps general notes separate from information about his family and finances. His test for a useful system is whether he can say, "Give Claude access to these scopes of my knowledge, but not everything." He left open where an employee's own expertise ends and an employer's information begins, a question that becomes harder when a personal agent encounters knowledge gained at work.

Davidson imagined a design in which his memory stays on his phone and a new service must request specific information. A car dealership's agent, for example, could ask for details, but the phone would seek his approval before s