Lovelytics: Multi-Agent Approach to LLM Task Automation for Business Users

Abstract

This paper addresses the challenge of automating business tasks using Large Language Models (LLMs) by focusing on two key aspects: generating high-quality prompts from unclear user input and executing tasks in a modular and scalable way. The system proposed combines DSPy (Declarative SelfImproving Programs)-driven prompt generation, which refines prompts based on feedback and task context, with a multiagent execution approach [1]. Unlike common industry practices, this system reduces manual effort by automating both prompt creation and task execution. The goal is to make AI-powered task automation accessible to non-technical users, allowing them to adopt LLMs into their daily workflow without the need for specialized knowledge. By democratizing task automation, the system opens up new possibilities for more efficient workflows across organizations.