Set the phone down, tap Start, and leave it.
One tap of Start opens the camera and the microphone (iOS asks once), records everything you say as a storyline, and watches the camera: whenever something new appears in front of the camera and holds still — a page with a question, a note, a sketch — it is read within a couple of seconds and answered from your storylines (a check also runs every 15 seconds). Nothing to tap until you are done.
Listen and Ask are extras for when you do not want a full session: Listen records only (camera off), Ask takes one spoken question without recording. Start covers both.
Talk to it with the word Lens: "Lens, what does the toll-keeper take?" asks a question (a camera frame goes along), "Lens, look at this" reads what is in front of the camera, "Lens, new chapter" starts a new storyline, "Lens, stop recording" ends the session. Those sentences are not written into the transcript.
Once, before you start: — every storyline is then mirrored to a "Lens Loop" folder on your Drive.
First, add your Claude API key in settings (the gear at the top right).