About
About Mnemoria
Mnemoria lets you see how an AI forms a picture of a person. As you talk, it shows what Claude, Anthropic's AI model, concludes about you, what it remembers, what it's given before it answers, and how its view of you changes. Every step is recorded, so you can inspect it, correct it or delete it.
Why it exists
AI systems increasingly remember things about the people who use them. What they keep, and how it shapes what they say next, usually happens behind the scenes. Mnemoria makes that process visible and testable, using its own memory pipeline built on Claude through Anthropic's API.
- Curious people can see what Claude concluded about them, which of their words each conclusion came from, and how that view changed over time.
- Students can watch each stage of the pipeline and work through seven short lessons on how AI memory works.
- Researchers can run controlled experiments on invented people in the Lab, compare settings, repeat runs, and export complete traces. The Methods page describes every rule and number.
Founder's vision behind Mnemoria
AI systems are becoming part of how people work, learn and make decisions. Yet very few people can see what happens behind the scenes: what a model concludes about them, what it remembers, and why its view of them changes.
Mnemoria makes that process visible, so anyone can see it, question it and correct it.
I do AI research myself, and I know how hard it is to observe and reproduce how a model forms and revises its picture of a person. So I built Mnemoria as a free, non-commercial tool that gives researchers, students and professionals a transparent way to study exactly that: inspectable traces, controlled experiments, reproducible settings and results they can export.
My hope is that understanding how AI sees us helps people use it more wisely, and helps research make it more trustworthy.
About the founder
Tunjay Seyidali is the founder of Mnemoria and a final-year BSc Computer Science student at King's College London. At King's, Tunjay is carrying out supervised AI safety research on how the moral judgments of large language models compare with the spread of human judgments.
Tunjay has held software engineering and data analyst internships, and has been a Youth Delegate at the UN ECOSOC Youth Forum (2025) and the World Bank Youth Summit (2026). Tunjay built Mnemoria independently, from the AI pipeline to the infrastructure and design.
Member of Claude for Startups.
Mnemoria is an independent project and is not affiliated with or endorsed by Anthropic. Claude is a trademark of Anthropic.
