Memory that belongs to you.

We're brothers. For months, we were each quietly building the same thing — a personal AI assistant, one apiece. We compared notes constantly, and kept arriving at the same question: what actually makes an assistant like this genuinely effective for the one person it works for?

Peter had spent fifteen years living the answer. Running communications for a company of more than a hundred thousand people, he worked beside an executive assistant who simply remembered everything — every preference, every past decision, every person in the room — and grew more useful every year because of it. That continuity, not raw intelligence, is what made her indispensable. Today's models are already remarkable; what they're missing is the part that knew you yesterday and will still know you tomorrow.

That part is memory. So we built it first — and we built it to belong to you, not to whichever AI tool happens to hold it this year. We're an unlikely pair to be writing software: Aaron is a lawyer with an MPA who co-authored a best-selling book on business ethics and teaches in the MPA program at a major university; Peter comes from film and medicine. But we both grew up writing code, and never really stopped. Aaron's instinct is accountability — what you're owed, and how you'd check it for yourself; Peter's is how the thing should feel to the person on the other end. A memory you own needs both.

MyPenny is the first thing we've shipped together: a secure, portable memory that any AI tool can read and write, owned by the person it describes. The longer goal is the assistant we each set out to build — one that gets more useful the longer it knows you. Memory is what makes that possible.

That's also why we publish a tamper-evident log of every access to your memory, answer the hard questions plainly on this site, and treat export and deletion as features, not support tickets. An assistant worth trusting starts with memory worth trusting.