This paper presents the ongoing experimental work related to the design of an AI assistant using API calls to an LLM to support educators in creating sound learning design (LD): LeDA. LeDA is one of the innovative features of a collaborative LD tool Balanced Design Planning, founded on the principles of student-centeredness and constructive alignment, that has so far been used by more than 2000 users in over 40 countries. We explain LeDA’s background in research and international projects, outline the software development, and present the results of ongoing validation. LeDA has been validated in 34 higher education courses, helping 22 teachers to generate, customize, and optimize LDs. Data on the use of LeDA has been automatically collected in the tool, user feedback gathered via integrated forms, and through a questionnaire administered among the teachers. In the current total of 664 interactions with LeDA, the rate of accepting AI’s suggestions without modification is around 40%, with the majority relating to creating teaching and learning activities (TLAs). User feedback suggests that teachers recognized LeDA’s potential in generating learning outcomes, topics, units, and TLAs, while they were less satisfied with the assistance in integrating innovative pedagogies. Generally, it was reported that LeDA is more efficient in generating solutions on a lower conceptual level, especially if it is provided with the user’s comprehensive input. The findings showed that LeDA was easy to use, but pointed to possible adjustments, which would contribute to higher usability.