This chapter covers
- Understanding how AI agents can be built using LLMs
- Creating a basic AI agent to demonstrate their value
Over the previous few chapters, we saw how large language models (LLMs) can assist us in testing. We also learned how to employ various prompt techniques to get the most out of LLMs and curate different prompts that can be utilized when required. This is a great position to be in, but what if we could take our new understanding one step further to create custom-made AI testing assistants?
As LLMs have advanced, so have the opportunities to create AI agents, applications that can take a goal and autonomously interact with other systems, collect data, analyze information, and adapt a suitable response to achieve said goal. In the field of AI, an agent can be implemented in many ways, but the goal tends to be the same—to create something that we can give a task to be solved. The scope of designing and building AI agents is large, but in this chapter, we’ll learn a bit more about their potential and how they work in the context of generative AI. We’ll also create our own basic test data AI agent to demonstrate the power and potential of this technology.