GAX: connecting assessment data to AI feedback and coaching
An AI architecture that grounds feedback and coaching in each person's real assessment results, without training models on client HR data.
NFN is building GAX, an AI architecture that links the individual, leadership, team and organization assessments of the GA Platform to AI feedback and AI coaching.
GAX does not fine-tune language models on client HR data. When a user starts AI feedback, the system structures that person's actual results, competency definitions, interpretation guidelines and score-based feedback criteria, and supplies them to the AI as context. The conversation stays within the platform's professional interpretation standards.
GAX also reconstructs the real competency hierarchy from the assessment's item groups and applies rules that stop the AI from inventing parent–child relationships that do not exist. In a general chatbot this would be a small error; in an assessment of a person's competencies it is a serious one. Different scales (5, 7 or 100 points) and different instruments for executives and team leaders are handled from the actual data.
The aim is to carry assessment results through feedback and GROW-model coaching into action plans and change management.
Original article: Money Today