Scroll to explore

Remember everything.

Omniscient AI to augment human cognition

Our vision is to endow humans with infinite memory.

Our goal is to build a world where humans can remember everything — every person they ever met, every conversation ever had, every place ever visited. All accessible with searchless, promptless recall powered by novel AI models of human memory.

Engramme helps you remember

Join the waitlist

Latest News

View All

Gabriel Kreiman gives a lecture in ISSAI summer school in Spain

Conference

October 2026

Beyond COLM Conference

We brought together AI researchers, engineers, founders, and friends from COLM and across the Bay Area for an evening of ideas, new connections, and great company.

Announcement

July 2026

Gabriel Kreiman gives a lecture in ISSAI summer school in Spain

Gabriel Kreiman describes the journey towards taking inspiration from neuroscience, cognitive science, and computer science to create Large Memory Models (LMMs) and build AI systems with human-like memory.

Publication

March 2026

Can machines imitate humans?

State-of-the-art AI can remarkably mimic human performance in complex vision and language tasks, based on a novel comparison method. This work has been accepted for publication in Nature Human Behaviour and has major implications for AI development.

Latest News

View All

Conference

October 2026

Beyond COLM Conference

Announcement

July 2026

Gabriel Kreiman gives a lecture in ISSAI summer school in Spain

Publication

March 2026

Can machines imitate humans?

Latest Research

View All

Memory is Not Search: Toward Proactive, Lifelong Memory in AI

This position paper argues that search is insufficient for memory and that AI systems need distinct mechanisms to proactively recall information from lifelong, multimodal memoromes — pointing to new datasets, benchmarks, and architectures as the way forward.

Predicting what people remember from movies

This work introduces machine learning models to predict whether specific events within movies will be remembered or not.

Stretching beyond the obvious

This paper introduces the Stretch-and-Squeeze (SnS) framework to map a neural unit's "invariance landscape," revealing how specific transformations like pose or lighting can change an input without altering the unit's response, thereby bridging the gap between artificial and biological visual perception.