Research &
evaluation
Evidence, model behavior, and the integrity of a conclusion.
Commission a focused review of AI responses, evaluation criteria, or a research question. Receive documented findings, limits, and practical next steps.
Independent research & applied intelligence
Research, evaluation, and systems shaped
by close attention to what actually happens.
Observe closely. Establish what holds. Put it to work.
Evidence, model behavior, and the integrity of a conclusion.
Commission a focused review of AI responses, evaluation criteria, or a research question. Receive documented findings, limits, and practical next steps.
The relationship between what is said and what people can do.
Commission a review of a user journey, support conversation, or sequence of messages. Identify friction and clarify what should happen next.
Practical instruments built around a defined need.
Explore a scoped prototype or review workflow. Establish the task, test its behavior, and document what is ready and what still needs work.
Independent investigation
A closer look at how five AI assistants examined and corrected their own answers—and the difference between acknowledging uncertainty and verifying a correction.
Explore the research →Local prototype · in development
A developing tool for reviewing customer-support conversations: examine the reply, identify what remains unresolved, and decide what to change or retest.
Explore two interactive examples of the review approach. The full application is in development.
About the project →AI behavior & evaluation research
→Suzanne’s UX LabExperience research & design
→Reply CheckupSupport conversation review · in development
→Standards of practice
Evidence before assertion. Clear scope before execution. Findings that disclose their limits.