Research notes from the lab.
Method notes, findings, and field observations.
PYX-Voice Leaderboard Refresh: What 13 New Models Tell Us About AI and Employee Surveys
Adding to the July 2026 leaderboard of seven frontier models, the September 2026 PYX-Voice leaderboard refresh tested 13 additional large language models on 84 employee survey tasks. For the first time, we can report an upward trend: nine models surpassed July's best score of 76%. The field is converging at the top, with nine models clustered between 79 and 83%, making model selection less about raw scores and more about which capabilities fit your specific workflow.
Read →
The Jagged Frontier: What Running LLMs on Employee Survey Data Taught Us
After publishing PYX-Voice, our team sat down to share what the experience actually taught us. The headline finding was one that surprised even seasoned practitioners: LLMs handled complex quantitative analysis more reliably than expected, while stumbling on what seemed like simpler scoping decisions. Performance was uneven in ways that defied prediction — and that unevenness is itself a signal worth paying attention to. This post captures what our subject matter experts and technologists are taking away, and where they think the field needs to go next.
Read →
The Human Impact of AI in Employee Listening
As AI takes on a larger role in employee listening, it raises questions that go beyond efficiency: what happens to the HR practitioners who built careers around connecting with employees, and to the employees whose sense of being heard shapes whether they keep speaking? This post explores the human impact of AI in employee experience — drawing on research from Stanford HAI, Self-Determination Theory, and employee voice science — and what it means to build AI that serves the humans on both sides of the listening process.
Read →
How to Know If AI Is Any Good at Employee Listening
AI in HR has no shared standard for evaluating quality — no bar exam, no FDA equivalent — leaving organizations to guess whether the tools they're deploying are actually reliable. This post introduces PYX Labs and the thinking behind PYX-Voice, the first benchmark built to close that gap in workplace AI: domain-specific, expert-graded, and designed to measure not just whether AI can complete employee listening tasks, but whether it applies the right judgment when it does.
Read →
What Frontier AI Gets Wrong About Employee Feedback: Inside the First PYX Labs Webcast
In our first webcast, Joe Freed and Sarah Foster sat down with Stanford's Melissa Valentine to unpack what PYX-Voice found when we tested seven frontier AI models on real employee-listening work. The short version: the models are more fluent than they are reliable, reasoning power didn't predict performance, and the harder and more human the feedback, the more they struggled. Here's the recap, plus the full recording.
Read →
AI Can Summarize Employee Feedback. A New Benchmark Shows It Doesn't Always Understand It.
PYX-Voice, the industry's first benchmark for how AI understands employee feedback, finds it often misses the meaning behind complex workplace experiences.
Read →