New Life Hack: Using LLMs and Constraint Solvers for Personal Logistics Tasks

a year ago

Using LLMs to generate constraint solver programs can help solve personal logistical tasks with multiple constraints.
The author faced a challenge assigning friends to escape rooms with various constraints like arrival times, departure times, and room preferences.
LLMs like ChatGPT and Claude 3.7 Sonnet initially failed to solve the problem but succeeded when asked to generate a constraint solver program.
Constraint solvers allow users to declaratively express constraints and let the solver explore possible solutions efficiently.
The author used Google's OR-Tools Python package to model the escape room assignment problem with hard and soft constraints.
Soft constraints included playing at least one room with each person and grouping specific friends together.
The solution involved tweaking time slots and constraints until a feasible schedule was found.
The author suggests that LLMs could be more effective if they integrated constraint solver libraries like OR-Tools.
This approach can be applied to other logistical problems requiring optimization under constraints.

Hasty Briefsbeta