As far as I can see, your question is related to TIMES in general, and not particularly to ANSWER-TIMES. However, I am afraid that this modeling issue may not be very clearly explained in the TIMES documentation.
In TIMES you can model the load curves in several ways. If you define the demand commodities at the DAYNITE level, the demand will be distributed according to the load curve over the different timeslices, but in this case you would have to define also all the technologies producing the demand commodity at the DAYNITE/WEEKLY or SEASON level. In other words, in this case you would basically allow optimizing the timeslice operation among the demand technologies. For example, if the demand represents lighting, some lighting technologies might end up producing "base-load lighting", while others might produce "peak-load lighting". However, you would also be able to restrict or eliminate such an optimization of the production by introducing additional FLO_FR constraints and/or timeslice-dependent availability factors at the technology level.
I myself think that such an optimization of the load distribution among the demand technologies would rarely make sense. Therefore, I would usually recommend defining the demand commodities at the ANNUAL level. In this case also all the technologies producing the demand commodity should be defined at the ANNUAL level, and the load curve will thus be taken into account at the technology level, while transforming the ANNUAL level output into the DAYNITE level inputs of each demand technology. In other words, in this approach you would not allow for optimizing the timeslice operation among the demand technologies, but all the demand technologies would have to produce the demand commodity according to the load curve. You would then also need to define a realistic annual utilization factor (NCAP_AF/NCAP_AFA) for all of the demand technologies.
I think these are the two main approaches for the modeling of the demand load curves. In both approaches you can additionally introduce e.g. night storage technologies (e.g. electric cars or electric heating with night storage) that may consume their input commodities with a different time distribution.
I hope this answer provides some help for choosing the appropriate modeling approach in your model.
Antti