Turning Large Language Models into AI Assistants for Startups Using Prompt Patterns

Abstract

Most startups operate with limited resources and experience. AI technologies enable them to accomplish many tasks under these constraints. The recent advance of large language models (LLMs) offers new opportunities to support startup endeavors. Given the nascent nature of LLMs, how they could be utilized to support startups is yet to be investigated. Since prompt engineering is believed to be at the core of the effective use of LLMs, we aim to understand how to apply prompt engineering to turn LLMs into AI assistants for startups. As the first step, we investigated the application of a set of prompt patterns to ChatGPT, arguably the most widely known LLM currently. The preliminary results show that some patterns are more suitable for brainstorming which is a typical activity conducted by early-stage startups. Prompt-tuned questions may lead to more specific and more detailed responses, but it is not guaranteed. Meantime, human factors play an important role in the effective application of prompt patterns. Large-size and systematic studies are needed to apply the right patterns to different questions, taking into account the differences among startups in terms of their startup knowledge, domain knowledge, and their attitudes and behaviors towards LLMs.