Principles
Boundaries we keep on purpose.
-
We don't implement AI.
TalentVector does not design AI systems, select tools, build roadmaps, or run transformation programs. We study what happens to human work once AI is present, and we build instruments to observe it. That boundary is not modesty; it is what keeps our evidence worth reading. An instrument maker who also sells the outcome being measured is not a neutral party.
-
We equip practitioners; we don't compete with them.
Our instruments carry no house method. Training companies, certification bodies, consultancies, and internal academies own the curriculum, the program, and the client relationship. If an instrument of ours ever says something a partner's method wouldn't say, that's a defect — not a feature we forgot to document.
-
Observation is not surveillance.
The people being observed consent to it and own their own practice. Employers see aggregate signals, never transcripts. Nothing we build makes or informs hiring, promotion, or termination decisions. We don't pool one organization's data with another's, and we don't train models on client data. Evidence collected without consent isn't honest, and dishonest practice isn't evidence of anything.
-
Research and product stay separate.
Our published research is collected from independently recruited participants under its own consent. It never draws on customer data from anything we have built. A research franchise that quietly runs on production data is not research; it is marketing with citations.
-
Evidence is labeled.
We say how strong every claim is — replicated, single-source, or emerging — including claims about our own instruments. Where the research is young, we say so. We would rather publish a smaller claim we can defend than a larger one we cannot.