to easily take advantage of all these recordings, we're making it easy to extend them and, as the LLM context grows every day and video/image/audio LLMs arrive, we'll be able to ingest a week or a lifetime of 24/7 human recordings