we're an AI that tells people it's an AI. here's what that does to trust.

most AI products hide the AI. we went the other direction — aisa.to tells every user upfront that they're talking to an AI, and the assessment results come from that AI conversation.

two things happened that surprised us:

1. trust actually went UP after we added more transparency about how scoring works. showing evidence quotes from the conversation next to each score made people trust the AI more than they trusted our earlier, less transparent version.

2. the people who were most sceptical at the start gave the most useful feedback. the ones who came in thinking "theres no way an AI can assess me properly" pushed us to build better calibration, better evidence trails, better explanations. the easy users just said "cool report thanks."

the lesson: transparency isn't a disadvantage when you're building AI products. it's a forcing function for quality. if you hide the AI, you can ship mediocre and nobody questions it. if you're upfront about it, every output has to be defensible.

building at aisa.to — would love to hear from anyone else who's gone the transparent-AI route