LLM personal assistant
Persona
202600 / Project overview
A mobile assistant that turns natural-language requests into practical actions across meetings, email drafts and social publishing workflows.
Routine work is split across calendars, email and publishing tools, while a chat response often stops at advice. Persona turns one natural-language request into controlled actions across those services.
01 / Challenge
The constraint that shaped the build.
The hard part was not generating text. It was translating an ambiguous request into a predictable action while coordinating an LLM, application APIs, stored user data and automation workflows.
02 / Approach
Make the rules visible.
I separated the conversational layer from the execution layer. The assistant interprets intent, while purpose-built services validate and carry out calendar, email and content tasks.
Natural-language meeting scheduling
Context-aware email response drafting
Social post generation
Service-based automation workflows
03 / Solution
The finished system.
A Flutter interface connects to Express.js and FastAPI services. Groq provides the language-model layer, Supabase stores application data and n8n coordinates multi-step automations.
Outcome
The final-year project demonstrates an assistant architecture where generated responses can lead to controlled, useful actions rather than stopping at chat.