Sunzu turns your customer data into synthetic humans your team can ask anything. Product, design and marketing teams run interviews, concept tests, usability studies and ad testing in seconds, not the weeks traditional research takes.
From your data to synthetic humans you can trust. The whole idea, in half a minute.
Synthetic user research means putting your questions to AI models of your customers instead of waiting weeks to recruit real ones. The models are built from your own customer data, and every answer traces back to a source.
Built from your CRM, interviews, surveys and analytics. Tested before use. Every answer cites its source.
An LLM asked to roleplay your customer. Fluent and agreeable, but not based on your data. That is not research.
| Dimension | Sunzu synthetic humans | Traditional research | Synthetic users with LLM |
|---|---|---|---|
| Time to answers | Seconds | Weeks | Seconds |
| Grounded in your data | Yes, built from your customer data | Yes, real participants | No, generic training data |
| Traceable and auditable | Every answer cited to a source | Partial, manual | None |
| Cost per study | Low | High (recruit and incentivise) | Low |
| Scale and reach | On demand, any segment | Limited by recruiting | Unlimited but ungrounded |
| Best used for | Fast, grounded first-pass research | High-stakes, definitive decisions | Nothing you need to trust |
Read the method: Engineered, not prompted and 9/10 teams make the same mistakes.
Transform your team. With Sunzu, anyone can ask anything and get answers in seconds. From your browser, or directly from Claude, Slack or Teams.
Turn 2 weeks of research into seconds. Launch customer interviews and concept tests, run usability and ad tests at lightning speed.
All managed by an AI moderator who designs and runs your tests, ensuring robust insights, real fast. So stop waiting and get insights.
When research is rationed, teams rely on guesswork. We build Sunzu for everyone, so anyone can ask anything at any time.
Teams become more customer-centred, and research stops being a bottleneck and starts being a habit.
Traditional panels are gamed: professional respondents hot-housed across studies, screeners cheated, and bots posing as humans. That rot is corrosive to data quality.
You already have real customer data. Sunzu turns it into synthetic humans you can rely on: research that holds up when the rot is everywhere else.
Prompting an LLM to roleplay a user is research theatre, not research. Generic AI averages toward an agreeable, frictionless person that does not exist.
Sunzu synthetic humans are grounded in cognitive architecture and behavioural economics. They satisfice, contradict themselves, and behave like your actual users do.
In just a couple of months, over 170 of our colleagues have run over 500 studies.
We’re now directly integrating synthetic humans into many of our research, product and marketing workflows.
In a market flooded with AI-generated noise, this was the first platform that genuinely earned our trust.
By grounding its models in research principles, behavioural science, and real customer data, it delivers insights that feel credible, transparent, and actionable.
It changed our perspective on what’s possible with AI in customer research.
We turn the customer data you already have into a synthetic audience your team can question in seconds.
Bring in what you already have: CRM exports, surveys, transcripts, NPS, product analytics, support themes, churn analyses, or research repositories. Sunzu maps each segment across motivations, objections, values, and decision drivers.
Your audience is mapped through our cognitive architecture, covering perception, memory, attention and goal-setting, so each synthetic human reasons like the customer it is modelled on. The full science is in the section below.
Launch a custom instance, or work straight from Claude CoWork, Anthropic's AI workspace. Run depth interviews, concept testing, UX testing and ad testing in one place. Or connect through MCP and launch depth interviews and concept tests from Claude, ChatGPT, or any tool that speaks it.
Everything in your research stack.
Depth interviews that surface the why behind your numbers, moderated end to end.
Put rough ideas in front of your audience before a line of code is written.
Rank the roadmap by what customers would actually use, not who shouts loudest.
Walk synthetic humans through your flows and see exactly where they hesitate.
Pressure-test copy and creative against the segment it is meant to move.
Open exploration or structured measurement, from the same synthetic audience.
Grounded in your data, not guesswork. Every finding traces back to evidence you already have.
Your synthetic humans start with your signal: customer records, research, behaviour, feedback, and business context. External enrichment supports the model, but does not replace your data.
Every answer is inspectable. Researchers can review why a synthetic respondent answered a certain way and what evidence informed the output.
Sunzu helps teams decide what to validate with humans, not avoid human research. It front-loads exploration and makes recruited studies sharper.
Run interviews, concept testing, UX testing, and ad testing inside a single audience workspace. No switching tools.
Sunzu connects to the tools where research work already lives: Figma, Jira, Confluence, and Miro. It speaks MCP, so studies launch from Claude or ChatGPT too. Something custom? Just ask.
Synthetic humans are only as good as the science underneath them. Sunzu is built on computational frameworks from cognitive science and behavioural economics, not generative AI roleplay.
We validate against real human behavioural data, not synthetic training sets. Our models overcome the three critical failures of current generative AI simulators: persona homogenisation, tunnel vision in isolated scenarios, and optimism bias that erases authentic friction.
The result: synthetic humans who behave like your actual users, including their frustrations, mistakes, and unexpected paths.
We model how people actually think, not just what they say. Our system simulates perception, memory formation, attention, and goal-setting using computational frameworks that mirror human mental processes.
This creates agents that reason like real users, with both deliberate analysis and fast intuitive judgments.
People don't optimise perfectly. They satisfice, rely on heuristics, and make irrational choices under pressure. Our decision model captures both rational evaluation and the computational constraints that lead humans to suboptimal decisions.
This lets synthetic humans predict not just ideal choices, but the messy, real-world decisions your actual users make.
Authentic behaviour emerges from context, history, and cross-scenario consistency. Unlike isolated response generators, our behavioural model maintains causal chains across situations.
It remembers past interactions, maintains preferences over time, and exhibits long-horizon patterns that reflect how real people navigate complex product experiences.
Every complex human behaviour decomposes into fundamental building blocks: basic psychological constructs, elemental responses, and atomic actions.
Our primitives preserve the heterogeneity of real users: individual differences, edge cases, and long-tail behaviours that get lost when generative AI averages toward a generic positive person.
Head of Commercial
Human-centred research and product, making synthetic audiences that enterprise teams trust and adopt.
Head of Product
Product strategy and engineering depth, turning hard questions into answers in seconds.
Head of Technology
Enterprise AI engineering, grounding synthetic humans in real behavioural data.
Book 30 minutes. Come with a live question your team is circling. Not a hypothetical. We will model the audience and run the first exploration together, so you leave with a clear read on whether Sunzu is right for what you are trying to do.
Book a callWe'll start with your need. 30 minutes, no deck, no commitment.
