An AI advisor with a stable identity — a name, a domain, a track record of prior positions, and a consistent point of view across sessions — rather than an anonymous model instance spun up per query. Named experts matter for accountability and for calibration: you can learn that one advisor is reliably too optimistic on timelines and another is right about pricing, which is impossible when every answer comes from the same undifferentiated assistant.
The identity also carries obligation. A named expert owns the answer it gives, defends it under challenge, and is confronted with it when the decision is reconvened.