Daily tool for
your market
research
Create your audience in seconds and start asking questions.
Test ideas, compare concepts, understand preferences, and uncover the reasons behind decisions using real-world data.
Ahead of a Swiss federal referendum, our synthetic population called the outcome correctly while published polling pointed the other way. The initiative was rejected, with no majority at either level. Read the write up →
How it works
Four steps, one sitting. Ask, meet your audience, place them in their world, read the result.
Ask your question
Describe the question you want to answer or the decision you need to make.
- 01 Which packaging design will customers choose?
- 02 Will users value this new feature?
- 03 How does this article change people's opinions?
- 04 Which message resonates most?
- 05 Why do customers prefer Brand A over Brand B?
Meet your synthetic audience
Your audience is made up of thousands of individuals synthesized from real-world data and your selected criteria. Each one responds from its own perspective, and you can open any single response to see the reasoning behind it.
You can build both a synthetic population and a digital twin of your own customer base. See how they differ →
"Design B looks like something I would trust for the kids. The other two feel like a promotion."
B"B stands out on a crowded shelf. I buy with my eyes when I am in a hurry."
B"I read the label first. A is the only one where I can actually find the ingredients."
A"C is cheaper looking, and that matters more to me right now than the design."
C4 of 10,000 responses shown
Each agent has its own media bubble
People do not form opinions in a vacuum. They are shaped by what they read, watch, and scroll past, and no two people see the same mix. Each agent gets its own media diet: which channels it uses, which outlets it leans on, how much weight each one carries.
We ingest real news and social content, tag every item by topic, sentiment, and audience fit, then match it to the agents likely to have seen it. Agents remember what they were exposed to, so they answer your question in that context.
The feed keeps updating, so the population stays current. When something happens in the real world, it moves opinion in the simulation the way it moves opinion in the market.
- Cut out sugary drinks for 2 weeks, energy is insane.Health
- Major brands commit to more recycled packaging by 2030.Packaging
- Snack idea: apple + peanut butter + dark chocolate chips.Snacking
Turn responses into decisions
Results land in a ready-to-read dashboard: ranked answers, segment breakdowns, and the patterns behind them. Drill into any segment, or go back and ask a follow-up question of the same audience.
See what a full report contains →Built for quantitative research
Structured, ranked, segment-level answers, at every stage from early concept work through to positioning.
Four questions teams stop skipping
These are decisions that get made with less evidence than anyone would like, because a full study takes longer than the question allows.
- 01Concept testing
Run new products, services, or features past your target audience before you launch. Reactions come back in minutes.
Outcome: Kill weak concepts before they reach the brief - 02Message testing
Test many copy or campaign variants at once. Find out which version performs before it goes in front of a real audience.
Outcome: Spend media budget on the line that already won - 03Pricing and positioning
Go beyond the last research cycle. Ask about motivations, trade-offs, and what would actually change someone's mind.
Outcome: A defensible price point, tested by segment - 04Market entry
See how a new market, region, or segment responds before you commit budget. Spot where demand holds up, and where it does not.
Outcome: Enter the two markets worth entering
What changes when the panel is simulated
Grouped by the constraint it removes, not by how the system is built.
Results in minutes, not weeks
Skip the fieldwork queue. Ask your question and see segment-level answers in minutes, instead of waiting on panel recruitment and scripting.
→ Research fits inside the decision windowAsk again, and again
Go back to the same synthetic audience with a follow-up question, a new concept variant, or a different price point, without commissioning a new study each time.
→ One budget line covers the whole explorationAny audience, including the ones you cannot recruit
High net worth individuals, C-suite buyers, niche regional segments. Groups that are expensive or impossible to reach through a panel are no harder to simulate than a general population.
→ Questions that were previously off the tableEvery market, one workflow
No separate translation cycles, no per-country feasibility checks, no difficult markets. Test across regions without running parallel studies.
→ One study, all your marketsNo length constraints
Traditional surveys have to stay short to keep respondents engaged. Ask as many questions as the research actually needs, without fatigue or drop-off.
→ Depth without penaltyNo fieldwork risk
No bot contamination, no speedsters, no straight-lining, no respondent fatigue. The population is calibrated once and stays consistent across runs.
→ Nothing to clean before you can read itSegment breakdowns by default
See how each demographic group responds without paying extra for cross-tabs or waiting for a second deliverable.
→ The cut you needed is already thereReproducible, not one-shot
Traditional fieldwork closes once the study ends. Rerun the same study as assumptions change, or as new data becomes available.
→ A living baseline instead of a dated reportWhere Foresenta sits
Against the two things teams currently use when a decision cannot wait.
| Traditional research | General-purpose assistants | Foresenta | |
|---|---|---|---|
| Time to results | Weeks to months | Seconds | Minutes |
| Cost per study | High: recruitment, fieldwork, project management | Low to none | A fraction of traditional research |
| Population grounding | Real respondents, but small samples | A single plausible-sounding answer, not tied to a real population | Calibrated synthetic population built from census and demographic data |
| Sample size | 100 to 2,000 people | One response, no distribution | 10,000+ simulated agents per run |
| Segment breakdown | Possible, but adds cost and time | Not available | Included by default, by demographic group |
| Reproducibility | Fixed once fieldwork closes | Answer shifts between sessions, no statistical basis | Rerun on demand against the same calibrated population |
| Hard-to-reach audiences | Expensive or impossible to recruit | No real respondents involved at all | Simulated directly, no recruitment needed |
| Output format | Structured report | Conversational text, not decision-ready | Structured, ranked, segment-level report |
The ones that come up first
Method, data handling, and what the agents cannot answer.
All questions →Not by prompting a language model to role-play a persona. We start from census and client data, build statistically weighted agent profiles, layer in media diet and context, then run a multi-step reasoning process on top. The population layer is math-grounded and traceable. The reasoning step is where interpretation happens, not the population itself.
We do not quote a fixed accuracy number, since it depends heavily on category and study design, and we would rather show you than tell you. The way to know is a parallel test: the same question fielded to a synthetic population and to a real panel, compared side by side on your own study.
As a public example, our synthetic population predicted a Swiss federal referendum result before the vote, whereas the official polls did not. See the full prediction →
No. Client data is used only for that client's own studies. It is never used to train shared systems and never shared across accounts.
Anything outside the scope of the data and context they were given for that study. We would rather be upfront about where that boundary sits than let a study appear to answer everything.
Run your first study this week
Bring a question you are already working on. We build the population, run the study, and walk you through the results. If you have a past study to compare against, we will run that in parallel so you can judge the output against what the panel told you.
Subscribe to updates
Method notes, study results, and product news.