For research, product and AI teams

Talk to your market before you bet on it.

Synthetic people with the real demographics of their country (age, education, income, work and household, drawn from official population statistics) and a psychology that stays the same from the first question to the last. Research, rehearse or test with them in hours instead of weeks, and see exactly where each one came from. Talk to one and see for yourself.

what happens before the words
Who she is
41 · Spain · two children · part-time office job · tight household budget · core fear: losing control
What she believes right now
belief.price_is_fair = 0.31
belief.brand_trust = 0.58
belief.needs_it = 0.64
So she decides to
intent = confront
goal = protect_self
And only then, says

"I like it, but at that price I'd wait for a discount. My budget doesn't stretch that far every month."

What it's for

When you need people to test with, and real ones are too slow.

Recruiting humans is expensive, takes weeks, and you cannot run the same person twice. These four jobs are what teams actually use the engine for.

Train conversational AI on real human dynamics

Generate dialogue that contains conflict, distrust, persuasion and repair — with the intent, goal and belief state labelled on every turn, computed before the text was written.

Find where your product breaks before your customers do

Put your system in front of users who escalate, contradict themselves, push back, or try to make it say something it shouldn't. Repeatably, and before real customers do it.

See Agent Check
Walk into the meeting with more than a hunch

Interview synthetic consumers with the real demographics of your market, in hours. Their profile stays fixed and explicit for the whole interview, with scale answers computed from it instead of improvised by the model, so the study with real people asks the right questions.

Walk into the conversation having already lived it

Practise a negotiation or a difficult decision against a counterpart with its own profile, its own priorities and no interest in agreeing with you. The first hard no comes from a synthetic person, not from your client.

The engine

A system prompt describes a person. This builds one.

Ask a general-purpose model to play a character and it will — but you cannot say where that character came from, why it answered what it answered, or how good the answer was. StrataSynth models the person explicitly, outside the text, so all three have an answer.

A psychological core, not a costume

Every synthetic person has identity, attachment style, core fears, biases and a life history. Reusable across conversations instead of thrown away after each run.

It decides before it speaks

Beliefs, relationship state and decision logic resolve first; language is rendered afterwards. On closed questions — scales and forced choices — the engine computes the answer from the profile and the model only puts it into words.

Structured output, not just text

Each turn carries the state the engine computed before the text — intent, goal, belief state and belief delta, relationship state — plus the communication act the generator reports. Written down, not inferred afterwards.

Quality measured without an LLM

Metrics are computed deterministically. No model grading its own homework — which is how circular evaluation hides problems instead of surfacing them.

Built on the engine

Three products already running on it.

The engine is not a demo looking for a use case. These are full products in production, each consuming the same platform through its API.

QualiSynthqualisynth.com

Qualitative research platform. Runs AI-moderated interviews with real and synthetic consumers, then turns full conversations into structured evidence — drivers, barriers, needs, and quotes — for agencies and insight teams.

qualisynth.com
Study modes
Human · Synthetic · Hybrid
Primary output
Drivers, barriers, needs, quotes
Built for
Agencies · Insight teams · Product research
ArenaSyntharenasynth.com

High-stakes practice platform. Professionals rehearse the conversations that matter — negotiations, objections, difficult decisions — against synthetic counterparts with their own profile, their own priorities and no interest in agreeing with you.

arenasynth.com
Session type
Live practice against a counterpart
Primary output
Rehearsal + feedback on what moved them
Built for
Sales · Negotiation · Professional training
Agent Checkagents.stratasynth.com

Test your AI agent with synthetic customers before real ones do. Customers from 8 countries talk to your customer service or sales bot in their own language, ask for what it must never grant, and you see the exact sentence where it slipped.

Try Agent Check free
Built for
Teams with a customer-facing AI agent
Primary output
Where it broke your rules, turn by turn
Start
Free trial: 3 conversations
Custom datasets

Tell us the scenario. We build the dataset.

Your scenario, your countries, your volume — generated by the same engine, delivered with the quality metrics and documentation you can see on the public datasets below. You are not buying more text: you get what was going on underneath, on every turn.

1
You specify

The scenario, the countries, how many people, how many conversations, and what you're training or testing.

2
We generate

Same engine, same deterministic quality checks that run on everything we publish openly.

3
You get it

The dataset with its metrics and a card documenting how it was made — auditable, not a black box.

A typical free dataset
  • • text
  • • speaker labels
  • • limited psychological consistency
  • • little or no explicit state
A StrataSynth dataset
  • • text + intent + goal + communication act
  • • belief state and belief delta
  • • relationship state and trajectory
  • • reproducible via seed and versioning
Sample datasetsFree · CC BY 4.0

See real output before you ask us anything

Public datasets generated with StrataSynth — belief tracking, relationship trajectories and ground truth labels included. Load them with datasets.load_dataset().

View org
New: v3StrataSynth Personas: 10,000 people from 8 countries

Spain, Mexico, the US, the UK, Germany, France, Italy and Brazil, each person in their own language, with a biography, a speaking style, Big Five and attachment scores and, new in v3, their money situation: whether they make ends meet and, in Spain, Mexico and the US, late payments, cards, savings and buy-now-pay-later. Ready to use as simulated customers for your agent.

From the blog

How the engine actually behaves

Sep 29, 2026
Synthetic persona datasets compared: which one fits your project?

PersonaHub, FinePersonas, PERSONA, Synthetic-Persona-Chat, Twin-2K-500 and StrataSynth Personas each solve a different problem. What each one offers, and when to pick it.

synthetic-personascomparison
Jul 6, 2026
Beyond Roleplay: Why Synthetic Personas Must Hold Identity Under Pressure

A comparative test between StrataSynth and two leading general-purpose LLMs, using a single senior persona pushed through twenty adversarial turns. The finding: the next frontier is not better text, but identity that holds under pressure.

evaluationidentity-under-pressure
Jun 4, 2026
Can Cultural Conditioning Survive Negotiation Pressure?

The same two synthetic negotiators close the same B2B deal 100 times — 50 in Great Britain, 50 in the United States — changing nothing but country conditioning. Some cultural signals weaken under pressure. Others strengthen. That asymmetry is the finding.

cross-culturalbenchmark
May 10, 2026
PsycheBench: an open benchmark for Synthetic Identity Engineering

Every system producing synthetic personas claims they are realistic. There has been no standard way to verify that claim. PsycheBench is an open evaluation suite: deterministic scoring, no LLM judges.

benchmarkevaluation
May 8, 2026
Synthetic Humans with Coherent Personalities: What Emerged When We Stopped Writing Scripts

We built Sofía Martínez Rojas without sales scripts or objection trees. In one negotiation session she held her ground under pressure: this is the record of that session.

emergenceidentity-under-pressure
Apr 19, 2026
Modeling Identity Under Pressure: A Step-by-Step Analysis of Synthetic Behavior

Most synthetic personas behave consistently — until they don't. A structured seven-phase analysis of coherence under sustained pressure.

belief-statecase-study
Apr 15, 2026
Synthetic Identity Engineering: a working definition

Most synthetic personas are costumes. Synthetic Identity Engineering is the practice of building the person underneath, and of measuring whether the identity holds instead of assuming it.

definitioncategory
Apr 11, 2026
What a Belief State Looks Like in a Synthetic Conversation

Most dialogue datasets give you the words. This explains the four belief dimensions recorded per turn — trust, hostility, self-worth, resolution — and why the state behind a turn is worth recording.

belief-statecognition
Apr 11, 2026
Why We Don't Use LLMs to Evaluate LLM-Generated Data

Evaluating AI-generated dialogue with another AI creates a circular system where both share the same failure modes. Here are the deterministic metrics we use instead.

evaluationmetrics
Apr 11, 2026
8,404 turns of psychologically grounded dialogue — now on HuggingFace

Four datasets where intent, goal, belief state and relationship dynamics were computed by the engine before the text was generated, not inferred afterward.

datasetshuggingface
Questions

The things people ask before the demo

How is this different from asking an LLM to role-play a persona?

The difference is traceability. Each person is sampled from published population data, and in generated datasets the engine decides intent, goal and belief state before each turn is written and records them as labels, so you can trace what was decided behind every line. On scales and forced choices, if you ask for it, the answer is computed from the profile rather than improvised. And whether a given setup holds under pressure is something we measure, turn by turn, not something we assume.

How do you measure quality without using an LLM as a judge?

The engine's evaluation metrics are computed with numpy, scikit-learn and sentence-transformers — never by a model grading output. A judge that shares the generator's blind spots hides problems instead of surfacing them. It also means we can measure whether a cheaper model degrades results, rather than having an opinion about it. Product-level quality verdicts, in the tools built on the engine, are a separate layer and are documented separately.

Which countries does it cover, and is the culture real or a stereotype?

Eight countries with population conditioning: Spain, the United States, the United Kingdom, Germany, France, Italy, Mexico and Brazil. It comes from official population statistics rather than word lists: each person's age, education, income, work situation and household, and in five of the countries their actual occupation, are drawn from that country's real distributions and stay consistent with each other. Changing the country changes who is in the sample.

Can I see real output before talking to anyone?

Yes, and without giving us an email. Public datasets are on Hugging Face under CC BY 4.0, and the live demo runs in the browser: talk to a synthetic person yourself and judge it.

Can I get a dataset built for my own scenario?

That is the main thing we sell. You specify the scenario, the countries, the number of people and the volume; it comes back with the same quality metrics and documentation as the public datasets. Tell us what you are trying to train or test.

Who owns the data, and are there real people in it?

The output is yours. There are no human subjects and no personal data in it — the people are synthetic, generated from population statistics rather than sampled from anyone. That removes the consent and data-protection conversation that a human panel requires.

Talk to us

Tell us what you need synthetic people for.

Custom datasets for your own scenario, a vertical built on the engine, or something we have not thought of yet. We would rather hear the problem first than sell you a plan.

• Custom datasets — your scenario, your countries, your volume
• The three products above, if one of them already fits
• Guidance on whether synthetic data actually helps your case
Provided by HexaForms