Who Owns Your AI Memory?

When we use AI tools at work, we’re usually focused on the output. What’s easy to miss is that the context behind that work – the reasoning, the iterations, the institutional knowledge – stays locked inside whichever platform you used. Over time, that adds up to a lot of your organization's intelligence sitting somewhere you don't control.

Austin Cai ’25 sees that as a problem worth solving. As Chief Growth Officer at XTrace, he's part of the team building what they describe as "private, portable memory infrastructure" for the AI era – a way for organizations to own and compound their intelligence instead of handing it over to the platforms they depend on.

The Problem

Current AI tools do remember things about you, but only within their own walls. ChatGPT has its memory. Claude has its own. Gemini has its own. None of them talk to each other, and none of it belongs to you.

"Built-in memory is a silo," Austin explains. "Your memory becomes locked inside OpenAI's walled garden, or Google's, or Anthropic's. If you don't own your memory, you don't own the agent."

XTrace is built around a different category of memory entirely – one they call belief revision system. While ChatGPT might remember that you prefer bullet points, XTrace captures why your team made a major pricing decision, what changed between version one and version three of a proposal, and what a customer said six months ago. That context is encrypted before it leaves your environment, travels across tools and teammates, and is owned entirely by you.

Think of it, Austin says, as "having pieces of your team's best thinking embedded inside your AI, carried forward across every tool and every session."

$3.3 Million and a Mission

XTrace recently closed a $3.3M pre-seed round led by Draper Associates. Two themes defined the raise: missionary conviction and mathematical trust. The team is building privacy as a foundation rather than a feature, using encrypted vector search so that even XTrace itself cannot access user data.

The analogy Austin reaches for is a familiar one: "XTrace stays relevant the same way Dropbox stayed relevant after Google Drive launched – by being the independent, portable, cross-platform option that works precisely because you're not locked in."

The capital is going toward product depth, distribution, and team. Their early traction spans startups, local governments, and regulated industries where data ownership and compliance stakes are highest.

For Fellow Founders

Austin’s advice to early-stage founders is characteristically direct and not uncommon: find a problem and solve it.

"As you build and solve more problems, you start to realize what actually matters for a startup and where your gaps are," he says. "That clarity helps you form the right team and pursue the right idea with conviction." He also makes no apologies about the role AI plays in getting there: "You can't fight AI adoption – you have to embrace it."

It's a fitting philosophy for someone building the infrastructure to make sure that when you do use it, you won’t lose it.