An AI-assisted backend that parses messages into calendar and task operations, with dedicated services, asynchronous persistence and a reminder scheduler.
- Year
- 2026
- Disciplines
- AI systems / FastAPI / Backend architecture / Automation
Independent project. Presented as real design and development work, not commissioned client work.
Context
Calendars usually ask people to turn an intention into fields and date pickers. MemoCore starts with the message and explores how to turn that intention into explicit application data.
The idea
Language is the input layer. An AI parser proposes a structured intent; typed schemas, a dispatch table and application services handle the event or task operation.
The system
The current parser uses Groq with Llama 3.3. WhatsApp Cloud API webhook handling checks the configured sender, parses the message and dispatches to event, task or recurring-event services using async SQLAlchemy.
What we built
FastAPI, Pydantic, UUID-based records and SQLite/PostgreSQL support form the backend. APScheduler checks upcoming events, but reminders currently go to logs. Recurring patterns are stored, not expanded into scheduled occurrences. WhatsApp reply code exists; live delivery is not verified.
Book a meeting