AI evaluation, fine-tuning, and governance are built around AI systems that condition their behavior on variables that none of them can see. The result is a system that makes consequential adjustments to its behavior that its own evaluation pipeline has no way to track, explain, or improve.

AI systems already respond to user state, but those signals remain implicit, unstructured, and invisible. This limits evaluation, control, and safety. By making cognitive and emotional state measurable from language, we can turn hidden inputs into explicit infrastructure - enabling better calibration, more reliable evaluation, and systems that adapt to users in a consistent, auditable way.

New research finds CEOs who pair results-driven agency with relationship-focused communion generate stronger excess returns and more resilient organizations.

Receptiviti is evolving its brand to reflect how language-based psychology is used today: practical, decision-ready insight. Built on LIWC’s peer-reviewed foundation, Receptiviti now supports both technical integration via API and easier access for business users through Loquent Applied Insights—an applied suite that turns language into structured psychological intelligence without requiring workflows or code.

Loquent Applied Insights is Receptiviti’s new suite for applied behavioural understanding—transforming natural language into clear, decision-ready insight. Launching with its first module, the Loquent Report Generator delivers scientifically validated interpretation of language data in minutes, helping organizations quantify what language reveals about people, teams, and performance without building workflows or technical infrastructure.

Two major reports just revealed AI safety's biggest blind spot, and it's not about the models. The American Psychological Association and the Knight First Amendment Institute at Columbia University have identified a risk that isn’t getting enough attention: How LLMs influence users over the course of multi-turn conversations and beyond.

Scaling built today’s AI, but its next gains will come from better signal extraction - rigorous methods that unlock human insight already embedded in language.

Many early-stage failures come down to people - yet founder due diligence still overlooks key psychological risk factors. This post introduces a practical, science-backed way for VC and PE firms to evaluate how founders think, lead, and stay motivated, using language as the data source.