Inspiration
Gig-work in the uk is becoming increasingly popular amid the cost of living crisis for people to supplement their income. Yet, there's no financial-driven solution to help plan gig work around busy lives.
What it does
Aggregates natural language prompts, bank statements and financial data to create an interactive work-scheduler that works around your calendar.
How we built it
Our frontend is built in REACT. Communicates with the backend using a FAST api. The backend is comprised of python scripts and sqlite database. The backend uses a constraint solver for reliable scheduling, and the LLM generates soft constraints from natural language to guide the scheduling. After this we use python to offer insights based on data.
Challenges we ran into
n8n doesnt let you run python! so we had to rebuild the system quite late into the project. Turning natural language into a standardised constraint schema is also very tough to design and scope.
Accomplishments that we're proud of
Managed to glue everything together last minute! Made a buttery smooth front-end. Cheeky constraint solver.
What we learned
Glueing stuff together is hard and system design is the goat.
What's next for Giggle
With more financial data we could make far more insights and use them in the constraint solver to make even more optimal timetable suggestions.
