Regarding privacy and security, you don’t collect data, but are you able to speak to whether you’re calling an API of an LLM? Those third parties could be collecting data. A formal privacy policy would be helpful there.

As far as product feedback, I am just not sure this is unique enough to be noteworthy. How is this different than having a simple ~one paragraph of custom instructions in a generic LLM system?

I think the real issue isn’t whether this is “just a prompt”.

It’s that venting itself is a dead end if the interaction has no friction, no constraint, and no consequence. An AI that endlessly validates you is comforting… once. Then it becomes noise.

That’s why I ended up building something orthogonal with https://pfff.me . Instead of chatting, complaints are compressed, constrained, and gamified. You don’t talk at an AI — you throw your frustration into a system that reacts, scores, distorts, or escalates it.

No productivity angle, no emotional coaching either. Just turning negativity into a mechanic rather than a conversation.

If an idea like AnnaAI feels interchangeable with a custom prompt, it’s usually because chat is doing too much of the work. Change the interaction model, and suddenly the question isn’t “which LLM is this using?” anymore.

Curious whether people here think catharsis alone is enough, or if friction is actually the missing ingredient.