You can programmatically add things to an AI Task prompt, but not in parallel like langchain does. If your knowledge base agent is enabled, all the relevant chunks it pulls are stored in

event.kb.results

. You can then iterate through this array.