Built on six years of gameplay and cognitive-science research involving more than 500,000 players, Signal evaluates how well AI models move real players toward a target mental state, and the scarce human-outcome signal behind it.
SAN FRANCISCO, CA / ACCESS Newswire / August 19, 2026 / Skillprint today launched the beta for Skillprint Signal Benchmark, the first public benchmark that scores how effectively AI models can enhance real human performance. Existing benchmarks measure what a model knows, how accurately it answers questions or how well it performs a fixed task. Signal measures something no one else does: how well a model adapts a live game, in real time, to move a real person toward a target mental state, such as greater focus, relaxation or improvement in a cognitive skill.
Signal places AI models inside live gameplay experiences with real participants. As someone plays, the model can adjust parameters of the game in real time toward a defined human outcome, such as greater focus or relaxation. Signal then measures the resulting change in the player’s state and ranks models based on how effectively they achieve that outcome.
Signal is built on six years of Skillprint research spanning more than 500,000 players and a library of more than 1,000 games, quizzes and simulations designed to understand relationships between gameplay, cognition, mood and player state.
The launch marks a new phase for Skillprint, which has spent six years building a gameplay-powered intelligence platform around a central belief: AI should augment human capability, not replace it.
“Most AI benchmarks tell you how smart a model is on its own. Based on our years of decoding human cognition at scale, we’ve built the one that tells you whether it can actually help a human being, and enhance their performance and intelligence. As AI becomes part of everyday life, we think that is one of the measures that will matter most,” said Chethan Ramachandran, Co-founder and CEO of Skillprint.
Measuring AI Through Live Human Interaction
Signal runs on genuine gameplay. During each session, an AI model receives information about the game and player behavior and adjusts game parameters in real time. Signal measures whether those interventions move the player toward a defined target state.
At launch, Signal runs on two instrumented games, Hextris and Box Tower, across target states including Relax and Focus. Skillprint is evaluating a growing roster of frontier and open-weight models, with results published to a public leaderboard as sufficient session data becomes available.
No single model currently performs best across every measured human outcome.
Rather than testing whether a model can produce a predetermined correct answer, Signal evaluates how models use visual input, memory and real-time gameplay information to respond as a person’s behavior changes. Over time, Skillprint plans to expand Signal beyond states such as Focus and Relaxation to evaluate whether AI systems can help people strengthen specific cognitive skills.
Because each evaluation is generated through a live human interaction rather than a fixed test set, Signal is designed to be more resistant to contamination, memorization and benchmark saturation. Every result reflects a new gameplay session and the interaction between a model, a game and a real person.
From Game Performance to Human Performance
Skillprint began developing its cognitive-science and gameplay foundation in 2020. Since then, the company has built a library of more than 1,000 games, quizzes and simulations, mapped more than 150 game mechanics to human behavior, and developed a framework spanning cognitive skills, moods and personality traits.
More than 500,000 people have played experiences used to train and refine Skillprint’s models. Each session helps the platform better understand how gameplay relates to the way people reason, adapt, learn and respond to changing conditions.
Signal builds on that foundation by making the human outcome itself part of AI evaluation. It’s built on live, consented, pseudonymous gameplay sessions. It is the closest available read of how a model changes real human behavior in the moment, and it cannot be scraped. Skillprint does not collect personally identifiable information. The platform analyzes gameplay behavior and produces structured signals related to cognitive performance and player state.
“We have spent years studying what gameplay can tell us about cognition and player state,” said George Kachergis, Cognitive Scientist at Skillprint. “Signal reverses that question. Instead of only asking what a person’s gameplay tells us about the person, we can ask what happens to that person when an AI system begins adapting the experience around them.”
Availability
Skillprint Signal Benchmark is available in beta today, with early results published as the leaderboard fills. AI teams and researchers that want early access to results and to the underlying human-outcome data can join the waitlist at https://skillprint.co/benchmark#waitlist.
Skillprint will continue to drive human-AI collaboration, helping AI Labs train models with unique data to work more effectively alongside humans to achieve their goals.
About Skillprint
Skillprint is a human-performance intelligence company using gameplay to understand and enhance how people think, feel and perform. Since 2020, Skillprint has combined cognitive science, game mechanics and AI to translate real gameplay into measurable human-performance signals. Skillprint’s mission is to help build a future where AI augments human capability. Learn more at skillprint.co.
MEDIA CONTACT
Kate Pietrelli
kate@pathadvisors.com
+1(760)-518-2633
SOURCE: Skillprint Inc.
View the original press release on ACCESS Newswire
Media gallery
