Original source text
TL;DR: ZetaChain and Verona are running the first study of what people want AI to remember, and what it should never keep. More than 300,000 people have joined Anuma, each with a private AI memory. Verona proves facts about you without exposing the data. The survey takes five minutes: take it here.ZetaChain is building the Private Memory Layer for AI: user-controlled memory, identity, permissions, and payments that work across AI models, apps, and agents. We believe what your AI knows about you should belong to you, with keys only you hold.
Today, we're excited to announce a research partnership with Verona, the intelligence layer for AI used by 150+ brands including Uber and Nike, and backed by $36M from Multicoin and Circle. Together we are running the first study of what people actually want AI to remember, starting with the more than 300,000 people who have joined Anuma since February. We are also exploring how verified context and private memory can work together in one product.
Memory versus verified contextZetaChain remembers what you tell it. Verona proves what's true about you.
Memory is what an AI keeps: encrypted, portable across models, and under your control. Verified context is what an AI can prove: a fact checked once at its source, with a proof you own that any agent you authorize can reuse without seeing the data.
An AI that remembers you is useful. An AI that can prove what it knows about you is trustworthy. Almost nobody is studying the gap between them.
300,000 people with private AI memoryAnuma launched in February. Since then, more than 320,000 accounts have been created, each one a self-custodial wallet, and Anuma users have made more than 1.3 million AI requests across 38 models spanning chat, image, and video. The live numbers are published openly on ZetaChain Research.
Every one of those accounts has an encrypted memory vault that only the user's key opens. We cannot read it, and neither can the model providers. Your writing style, preferences, projects, and files follow you from one model to the next without starting over, and you decide what stays. Since late September that same memory also powers Nearby, which introduces Anuma users to people near them worth meeting. What an AI remembers now shapes who you meet, not only what it answers.
Why VeronaVerona makes user-verified data portable, private, and programmable. Its Truth Engine uses zero-knowledge proofs across websites, email, passports, and apps to turn a claim into a provable fact that the user owns through a Meta Account and any agent they authorize can reuse. Verify once, reuse everywhere, expose nothing.
More than 150 brands, including Uber and Nike, already build on it, and Verona is backed by $36M from investors including Multicoin and Circle.
Verona is the natural partner for this question because it sits on the other side of the line we want to study. We hold what an AI remembers. Verona holds what an AI can prove. The research asks whether people see that line at all, and where they want it drawn.
The research: What People Actually Want AI to RememberThe question is simple to ask and hard to answer. What do people want an AI to remember about them, and what should it never be allowed to keep? Underneath it sit the questions that decide whether AI memory is trusted: who decides, how you find out what is stored, and what happens when you revoke it. And now that an AI can introduce people to each other, a newer question sits beside them: what would you let an AI share about you to be discoverable, and what should it keep private?
The study covers six themes:
What people want AI to remember
What AI should never retain
Memory portability between models and apps
User visibility, permission, and revocation
The difference between remembered information and verified context
What people would let an AI use to introduce them to someone, and what stays private
We are asking Anuma users, Verona's community, and the broader AI user community. ZetaChain drafted the scope and questions from what we see in Anuma, and both teams reviewed them. Both teams will review the findings together and publish a report in November. Responses are anonymous and reported only in aggregate, and the full question set will be published with the report.
Take partThe survey is anonymous, needs no account, and takes about five minutes. It closes on October 27. Take it here: Take the survey.
Follow @ZetaChain and @verona_dev for the findings, and join the discussion on Discord.
What comes nextThe survey opens today and the report publishes in November, followed by an X Space with both teams and a series of data cards. We are also exploring how Verona's verified context and ZetaChain's private memory work together, so an agent can act on a fact about you without ever holding the data underneath.
About VeronaVerona is the intelligence layer for AI: a network where facts about the real world are proven once, owned by the user, and reusable by any agent they authorize, with the data underneath never exposed. Verona is used by 150+ brands including Uber and Nike and backed by $36M from investors including Multicoin and Circle. Learn more at verona.dev and in the Verona docs.
About ZetaChainZetaChain is building the application layer for private AI: user-controlled memory, identity, permissions, and payments that work across AI models, apps, and agents. Its first consumer product is Anuma, a private multi-model AI app with one memory across every model, joined by more than 300,000 people since February. The next phase is bringing that application layer to Solana and opening it to a broader ecosystem of AI apps and agents.
Follow ZetaChain on X (Twitter) and join the conversation on Discord and Telegram.
Categories