Science | Synthetic Users

A parity score is only as good as the test behind it. Five layers make a Synthetic Organic Parity test hard to argue with. Here they are, in one simple healthcare example.
Before polls opened in Brazil and the US, synthetic users made public calls on every race. Here is how they did, wins and misses.
Most synthetic-human systems start from a description of a person. We think the more interesting question is what happens when you start from their behavior.
A language model wakes into every question with no memory of the last one, so everything that feels like memory is a note someone handed it. What it takes to make those notes trustworthy - and why the one that a model writes is never allowed to say anything new.
Until now, a client's documents shaped who our Synthetic Users were. Now they are also in the room during the interview, consulted question by question, with a record of what grounded every answer.
Synthetic people are already being used to forecast real survey numbers. Here is what those forecasts actually estimate, how to score them against a bar that is achievable, and why the honest bar is never zero.
Every Synthetic User in your concept tests and user tests now shows you where their attention lands, on the package, the price, the headline, the button, conditioned on who they are and what they told you. Validated against thousands of images of real human eye-tracking. Nobody else does this.
Why we built our own platform to version, evaluate, and continuously improve the prompts behind our product and how a loop of automated evals and human-approved AI proposals replaced grading on gut feel.
Language models are built to outgrow us. Synthetic Users is built to stay exactly as human as you are and a new result that reads whole sentences out of a brain scanner is why we keep walking toward the source instead of away from it.
Synthetic Users is an extraordinary qualitative instrument. It's fast, deep, and refreshingly honest about what it knows. Understand how it produces a ranking and you will pull cleaner, more trustworthy answers from it than from almost anything else in your stack. Here is how to read the numbers like a pro.
Everyone else is wiring models together to chase superintelligence. We wire them together to rebuild something rarer: a human one. And like a real nervous system, it doesn't start at the cortex. It starts at the core.
When the shapes inside language models started predicting the shapes inside human brains, and what that means for predicting people.
Synthetic Users without calibration are individually believable, but collectively wrong. The missing piece is calibration, not better models.
Most research asks what people say. The problem is people don't do what they say. This piece breaks down the gap between stated and revealed preference — and why behavioral modeling, not better interviews, is how you close it.
Iris, what is the difference of using agents to accelerate research.
You don’t need a twin for “a parent in rural Ohio who shops weekly at Walmart, prefers fragrance-free, and has a toddler with eczema.” You sample a parent profile with relevant traits and constraints, add retail and dermatology context, and generate behaviors consistent with both.
LLM-powered Synthetic Users have crossed from concept to validated method. This proves they can predict human behavior accurately, letting teams run fast, low-cost behavioral experiments without replacing real participants.
Gartner’s latest report on AI-powered synthetic user research cites Synthetic Users as a leader.
Shuffle v2 is a feature that intelligently shuffles between multiple large language models via a routing agent to produce more realistic, diverse Synthetic Users with better organic parity.
Synthetic Users use a “chain-of-feeling” approach—combining emotional states with OCEAN personality traits—to produce more human-like, realistic user responses and yield richer UX insights.
A paper that thoroughly executes a parity study between Synthetic and Organic users.
Here is a compilation of all the papers that help make a case for Synthetic Users.
Synthetic Users balances aligned and unaligned models to maintain diversity and authenticity in simulated users while ensuring ethical standards and user expectations are met.
Synthetic Users and AI are transforming research methodologies, offering innovative, cost-effective alternatives to traditional human subject studies.
When we compare different studies, especially looking at what synthetic (artificial interviews) and organic (real-world interviews) data tell us, we often find they mostly talk about the same things but there's also a bit where they don't match up. This gap is super interesting because it's like finding hidden treasure in what we thought we knew versus what we might have missed.
An LLM on its own is a biased narrator. Our job is to know exactly how.
The transition towards Continuous Insight™ aligns research activities more closely with the dynamic needs of the business and ensures that product development is continuously informed by up-to-date user insights.
Large language models (LLMs) like GPT-4 serve as powerful "vibes engines," empathizing with diverse groups and generating contextually relevant content. Their applications span market research, customer support, user experience design, and mental health support, offering invaluable insights and personalized experiences. While not infallible sources of truth, LLMs enable creativity, personalization, and connection within the realm of human language.
How Synthetic Users is changing the research process.
How do we know we are right? How do we know our Synthetic Users are as real as organic users? We compare.
Synthetic Users are evolving to address criticism about their generalist nature by incorporating representative data sets and personal narratives. Updated for 2026: calibrated populations, brain-grounded profiles and live quant.
Determine your interview target for achieving top