Oooh this is exciting! so when an agent pulls health data e.g. from Apple, is it raw or will it be unified into a standardised format before analysis? or does the agent handle device-specific formats itself?