DEV Community

Jubilee Nwigwe
Jubilee Nwigwe

Posted on Originally published at Medium

Giving a study chatbot long-term memory with Django and Walrus Memory

Most chatbots forget you when the conversation ends. For the Walrus Sessions 8 hackathon I built Study Buddy, a Django chatbot for university students that remembers each student between sessions using Walrus Memory.
How it works: every student gets their own memory namespace. Each message is saved as a memory. Before every reply, the bot recalls the most relevant memories and gives them to the model, so it can say things like "we were waiting for your supervisor's decision, how did it go?" without the student repeating anything.
Three classmates tested the live app with 10+ messages each. The hardest parts weren't the memory: the free Gemini tier allowed just 20 requests a day, so I added automatic fallback across models.
The full write-up, with code and screenshots, is on Medium:

https://medium.com/@jubileenwigwe001/how-i-built-a-study-chatbot-that-remembers-each-student-between-sessions-django-walrus-memory-5f125e5e39e4

Code: https://github.com/Nwigwe-Light/study-buddy-walrus-memory

Live demo: https://study-buddy-walrus-memory.onrender.com

WalrusMemory

Top comments (0)