Comment by aregnzsdejan

Love this. I’ve had the same issue using LLMs to sanity-check product decisions - they’re too agreeable if you give them any context that reflects your own thinking. Also even if you ask them to be brutal they sort of always come up with some bs Hail Mary positive spin. I've been playing around with this for best in practice QC doc (ironically) and the internal debate is pretty helpful. I will keep you posted on how it goes though. Would be good to be able to add more than 2 personas though...