Insight
Static Personalization Was the Best We Could Do
Role-based dashboards weren't laziness — they were the ceiling. Software could know a user's role, entered once at setup. It couldn't know their situation, because reading the situation takes judgment, not fields.
The only signal we had: identity
Role, persona, permission — the inputs to personalization, captured once at setup and frozen thereafter.
The situation was illegible
A messy account, an ambiguous email, a carrier's shifting appetite — language and judgment, not fields. Unreadable to old software.
The situation became legible
Once a system can read "a change request hidden inside a friendly email," it can compose for that moment instead of guessing from your role.
It's worth being fair to the old approach. Role-based dashboards, persona defaults, and permission-based views weren't laziness — they were the ceiling. I built plenty of them. Software knew a user's role because someone typed it in at setup. It couldn't understand their situation, because that took judgment, not data fields.
So products personalized around identity: role, persona, or permissions. One interface for each type of user, unchanged until someone updated the settings.
What changed isn't that we care more about users. It's that software can now read context. It can notice that an email contains a mid-term policy change, or that a submission needs extra review, and adapt the experience while it's happening. Not always right, but readable.
Don't over-credit the AI. Credit the new signal: the situation is finally readable.