The mistake almost everyone makes
The quickest way to put an AI assistant in a mobile app is to embed the credential in the app and let the phone call the AI provider directly. It works on the first try, it ships in an afternoon, and it means the key is now sitting inside an artefact that anyone can download and take apart. Keys extracted this way get abused at someone else's expense, usually for weeks before anyone notices.
CHiiRO does not do this. Your phone talks to CHiiRO; CHiiRO talks to the assistant. The credential never leaves our side and there is nothing in the app to extract.
Your questions stay out of the logs
There is a second, quieter decision here. Application request logs are the most-copied, least-guarded data any service has — they get shipped to monitoring tools, sampled into dashboards and kept far longer than anyone intends.
The assistant's path is deliberately placed so that what you type never enters that pipeline, and there is an automated test that fails the build if it ever starts to. The point isn't the test; it's that the decision was made early enough to be testable at all, rather than being a cleanup task nobody gets to.
What it deliberately isn't
CHiiRO has not enabled conversational characters, because that is regulated territory and the questions there deserve a proper answer rather than a fast one. The distinction between an assistant that helps you use a network and a synthetic person who talks to you is a real line, and we are on the near side of it on purpose.
It is not a persona and it does not pretend to be a person.
It is not a mentor, an advisor or a coach, and it will not be marketed as one.
It does not read your messages.
It is not a conversational character.
Honest about the shipping build
The server-side design is in place and deployed. The app that reaches that design is not yet on the public release track — a build in the pipeline is where the client half of this lands. We're saying so here because "your key never leaves our servers" is a security claim, and a security claim that is half-shipped has to be described accurately or not made at all.
When the release ships, this paragraph changes and the change is dated on the disclosures page.
What you get
What you ask the assistant doesn't end up in a request log.
Get into the right conversation faster, without learning the product's vocabulary first.
An assistant that doesn't pretend to be a person.
Questions and answers
Q.Does CHiiRO have an AI assistant?
Yes. CHiiRO has an in-app assistant you can ask questions in ordinary language. It is brokered server-side, so the AI credential is never present on your device, and the path it uses is kept out of the application request logs.
Q.Is my CHiiRO assistant conversation logged?
The assistant's request path is deliberately placed outside the application's request logging, and an automated test fails the build if that ever changes. CHiiRO made that decision early enough for it to be enforceable rather than treating it as later cleanup.
Q.Can I talk to an AI mentor on CHiiRO?
No. The assistant helps you use the network; it is not a persona, a mentor or a coach and does not present itself as a person. CHiiRO has not enabled conversational AI characters, because the age-rating and regulatory questions there need answering first.
Q.Why does it matter where an AI assistant runs?
Because an app that calls an AI provider directly must carry the credential inside the app, where anyone who downloads it can extract it. Brokering server-side means there is nothing in the app to take. It is a common shortcut and a consequential one.