Missing Information Is Part of the Business Problem
In Experiment #001, most inquiries were incomplete, and the useful step was making the gaps visible before acting.
- Date
- 15 Aug 2026
- Topic
- customer workflows
- Source experiment
- #001 — Customer Inquiry → Next Action
Of ten synthetic customer inquiries, most lacked at least one detail needed to act. Intent and urgency were not always visible from the wording, and a price question did not reliably mean the customer was ready to buy. The human decided quickly and intuitively; the reasoning was not consistent from case to case. When the AI was asked to separate what was known from what was unknown before proposing an action, the resulting actions were more defensible and the structure repeated cleanly across cases. The reading we take is that missing information is not noise around the inquiry — it is part of the inquiry. Handling it explicitly is the work. Ten synthetic cases, one human, no timing. An observation, not a law.
Evidence this rests on
- Most of the ten inquiries lacked at least one detail needed to quote or act.
- A price question did not reliably indicate high buying intent.
- AI output was more useful when it distinguished known from unknown before proposing an action.
- Human reasoning was intuitive and not standardized across cases.
Related findings
Research updates
Follow the research
Get new FIELD experiments and findings when we publish them.