Carry the work forward.
Decisions and unfinished questions should survive the session that produced them.
Why HOM exists
AI can be brilliant. It should also remember.
HOM began with a simple frustration: meaningful work should not have to restart every time a conversation, model, or provider changes.
The problem
We keep teaching AI the same history.
People repeat decisions, preferences, constraints, and hard-won context because most AI relationships end at the edge of a session.
HOM is being built so the useful history can continue—without making one model provider the permanent home of that history.
What guides us
Continuity with human control.
The experience should feel natural to a user and remain defensible under technical scrutiny.
Keep intelligence replaceable.
A better model should be an upgrade, not a reason to rebuild the relationship from zero.
Make important change visible.
When remembered information changes, the system should be able to explain that change.
Building HOM
Research-led. Product-minded.
HOM is a focused product and research effort, shaped by open-source engineering, academic literature, and dialogue with the research community.
DEVELOPMENT & RESEARCH
Walid Saidi
Developer · Independent researcher
Walid develops HOM and leads its independent research. Academically trained as an engineer, his work focuses on persistent AI memory and accountable cognitive systems.
H
House of Memory
A place intelligence can return to.
The name is architectural on purpose. HOM is becoming the working environment; AIMOS is the memory system beneath it. Together, they are designed to feel less like another disposable conversation and more like a place where work can grow.
Built openly
Useful to people. Inspectable by experts.
The product story belongs here. The architecture, benchmark protocol, papers, and evidence belong in the developer and research sections.