Intellectual Frontiers Press Fieldbooks

You Are Not the User

How B2B Software Lost Touch With the People Who Do the Work, and How AI Gives Us a Chance to Fix It

If you want to know how users use the software, watch them use it.

A nurse says, “I need to know which patients need my attention.” Support, Customer Success, Product, Design, Engineering, and QA each do competent work on that sentence. Months later the nurse gets a status column and says, “This doesn’t help me decide who to call.” Nobody failed. The system worked the way it was designed to work, and the design is the problem.

Forthcoming. The book is in review; its companion and tools are open now.

Competent, caring people can build the wrong software when the information reaching them has passed through a chain of handoffs.

B2B software feels clunky because the people who know the work could never build the software, so the industry built a chain of translators between them, and every translation lost something. This fieldbook treats every internal function as a sensor that reports evidence about users, never a stand-in that speaks as the user, and gives teams the tools to tell the difference.

AI is weakening the constraint that made the chain necessary. Used well, it shortens the distance between the person doing the work and the software. Used badly, it repeats the old mistake faster.

If you want to know how users use the software, watch users use the software. Everything else is evidence at a distance.

What you will be able to do

• Label every claim about users with how strong the evidence is and how close it came to real use
• Tell QA from user acceptance testing, and simulated UAT from the real thing
• Turn complaints into testable hypotheses instead of feature requests
• Watch work instead of collecting opinions about it
• Weigh a workflow by frequency, friction, economics, and clinical consequence, not clicks
• Design when physicians, nurses, billing, buyers, and patients legitimately want different things
• Use AI to put working alternatives in front of real users instead of writing tickets faster

The premise

A nurse says, “I need to know which patients need my attention.” Support, Customer Success, Product, Design, Engineering, and QA each do competent work on that sentence. Months later the nurse gets a status column and says, “This doesn’t help me decide who to call.” Nobody failed. The system worked as designed, and that is the problem.