Helping AI agents search to get the best results out of large language models

EnCompass executes AI agent programs by backtracking and making multiple attempts to find the best set of outputs generated by a large language model (LLM). This approach could help coders work with AI agents more efficiently.