What I do... Err, what my internal stated policy is is to use rationality mainly on meta level in deciding which intuitions, authorities, algorithms, or other sources to trust how much, as well as for tuning the knobs on intuitions, either directly or by framing, and setting up environments that will make sure the right decision procedure is used. For actual decisions either intuition or numbers are supposed to be used.

In practice poor mental health, akrasia, and the general unreliability of the human brain makes such a fragile setup break frequently, but i haven't been able to come up with anything better. Oh, and because I kind of suck at doing anything like that in the first place. So, yea... But it might work out better for someone with a less sucky brain.