playboox principles
The homepage says 100% human decisions. That is not a slogan. It is a mechanism. These six rules are how the product is built. Each one states how it is enforced, so you can hold us to it.
Assistants draft summaries, suggest questions, and propose next steps. That is the whole job. No candidate outcome is ever produced by a model.
How it's enforced Decision steps in every process are human-only by construction. There is no configuration that lets an AI advance or reject a person.
Advancing or rejecting a candidate requires a named person and their written reason, captured at the moment of decision. Not a batch action. Not a default.
How it's enforced The decision form requires a rationale. The record stores who, when, and why.
The process you design is the process that runs. Steps happen in order, for every candidate, every time. A busy week does not create a shortcut.
How it's enforced The process engine refuses out-of-order transitions. There is no bypass.
Every step, submission, and decision lands in an append-only record you can read back. When someone asks why, you have the answer.
How it's enforced Records are written at execution time, not reconstructed later.
Candidates talk to their own assistant. The employer receives only what the candidate submits: answers and documents. Never the conversation.
How it's enforced The employer-facing surface can only read submitted answers and documents. Candidate chat content never leaves the candidate's assistant. Candidate data is never sold and never used to train models.
Every applicant to a role answers the same questions and is reviewed on the same brief format. People are compared on substance, not on polish or luck.
How it's enforced Application questions are fixed per role version. Briefs are generated in one format for everyone.
These rules are how the product is built, not a policy we promise to follow. If you find a place where the product violates one of them, write to galem@automi.ai and we will fix the product.