Pega GenAI brings more LLMs to low-code automation workflows

“Embedding AI tools into applications platforms is a trend that came quite naturally out of generative AI’s strengths and also its weaknesses,” he said, adding that an example of the former is that LLMs are “particularly useful when working with a specific context. The application platform knows the context of the user, the task, and the broader workflow.”

As a result, he said, the “augmentation and automation capabilities can be applied to that very targeted context. One common use case is a targeted, integrated, conversational virtual assistant, sometimes called a copilot.”

As for model choices, Elliot said “different tasks are performed better and worse with different language models. Also, companies may have existing preferences for which language models they prefer. This may be based on performance, pretraining, or cost.”