Rayhan Mendis
← All work

Side project · 2023 · Solo build

Mood: an AI journalling app

A journal that reads each entry, works out the mood behind it, and charts how you've been feeling over time.

Screenshot of the Mood journal

The problem

Journalling is useful, but it's hard to see patterns across weeks of entries. I wanted a journal that did that reflection for me.

How it fits together

  1. 01Journal entry
  2. 02LangChain + OpenAI
  3. 03Zod-validated analysis
  4. 04PostgreSQL
  5. 05Mood chart

What I did

  • Built the app with Next.js, TypeScript and Tailwind, with Clerk for sign-in, deployed on Vercel.
  • Used LangChain and OpenAI to analyse the sentiment and themes of each entry.
  • Validated the model's structured output with Zod before saving it with Prisma.
  • Charted mood over time so patterns are visible at a glance.

Stack

  • Next.js
  • TypeScript
  • LangChain
  • OpenAI
  • Zod
  • Prisma
  • Clerk
  • Tailwind CSS
  • Vercel