Penn student launches AI platform designed to remember user conversations, identify patterns

A Penn undergraduate student recently launched their artificial intelligence assistant designed to remember users across multiple conversations.

Engineering junior Jonathan Wallace created and launched Engram, an AI assistant that builds a persistent model of a user’s goals, projects, and thinking patterns. According to Wallace, the platform — which went live on March 28 — has already amassed over 1,000 active users.

In an interview with The Daily Pennsylvanian, Wallace explained that Engram aims to address a core limitation in existing AI tools — their inability to retain and meaningfully use long-term user context. 

“They’re really, really good general assistants, but if a person was trying to get assistance with writing, they have to explain their style over and over,” Wallace said.

Wallace initially developed Engram as a personal tool before expanding it into a broader product. According to him, the platform is designed not just to store past interactions, but to build a deeper understanding of how a user thinks.

“The core difference is what this does with the memory,” he said. “With most assistants, memory is just like a sticky notepad.”

Unlike existing systems that primarily log past conversations, Wallace said Engram uses memory to identify patterns across a user’s work and goals, allowing it to generate more personalized responses and even initiate conversations.

“We track things like the Big Five personality traits as probability distributions, a nine-factor semantic memory system, and a personalized intelligence briefing around your priorities every morning,” Wallace said.

Wallace explained that most of Engram’s current users are students who primarily use the tool for academic purposes. Some use cases, he added, have extended beyond his initial expectations.

“I have heard from a particular user that they were using it in the context of philosophy to understand the relationship between human cognition and AI,” Wallace told the DP.

Because Engram’s core functionality relies on long-term memory, Wallace also said privacy and user control were central to the platform’s design. 

“If I want to delete any memories that the AI makes, I can do that easily and painlessly,” Wallace said. “I designed a complete memory dashboard where you can remove any memories you want to, and once it’s gone, it’s not retained anywhere.”

Wallace told the DP that he considered the broader, industry-wide shift toward more autonomous AI systems while building the platform. While the assistant can initiate conversations without prompts, he said he has taken a cautious approach to expanding its capabilities.

“I want it to be done thoughtfully; I want it to be done carefully; and I want it to be done with the user’s privacy in mind,” Wallace said.

Wallace initially built the system on a personal laptop. He recalled having to “be a little bit scrappy” and “not waste any resources” when developing the initial product.

“It does help keep the system as lean as possible in a lot of regards,” Wallace reflected.

Wallace explained how support from Penn’s innovation ecosystem helped benefit the project’s success. An award from the Penn Engineering Innovation Fund, he said, helped fund early market research and infrastructure for the program. 

“It was a huge honor,” he said. “It really helped validate the idea as a sign that it was a good one.”

Wallace credited Penn’s interdisciplinary environment with shaping the development of the technology, pointing to both academic and extracurricular influences — including conversations within his a cappella group and mentorship from his philosophy professor.

In the future, Wallace said he hopes to increase Engram’s user base and expand its capabilities while maintaining its personal focus.

“In the next year, I would hope to continue to scale and bring the promise of personal AI to as many people as possible,” Wallace said. “We haven’t started charging a single dollar yet; that’s intentional. We wanted to prove the product first, and we’re ready to flip the switch.”

Wallace explained that he hopes Engram contributes to a broader shift in the artificial intelligence industry toward more user-centered design.

“That’s what it should be about,” he said. “Refocusing and reframing the field of AI and the industry of AI to become more about the user.”

---