Company profile · 4 min read
Artificial Societies: simulated audiences for consequential decisions
Artificial Societies combines behavioral science and AI audience simulations for marketing, communications, reputation, investor and strategy decisions.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Artificial Societies is a market-research and strategic-communications product for teams that want to test a decision against a simulated audience before spending money or taking it public. Its practical question is not whether an AI persona sounds human; it is whether the simulation adds useful directional evidence alongside the research a company already trusts.
What it does
The current site describes networks of personas and methods including mix-method surveys, interactive focus groups, interviews, conjoint analysis and scenario cascades. It says teams can explore how narratives shift the opinions of customers, investors, regulators and opinion leaders, and lists solutions for reputation, crisis communications, product positioning, investor communications and corporate strategy. The case-study page shows the intended enterprise workflow, including a Teneo example modeling more than 180,000 perspectives.
The product evolved from the founders’ earlier Reach experiment into Radiant, a bespoke enterprise simulation service. The YC launch says the company has built more than 2.5 million AI personas and delivered more than 18 million responses to global enterprises. Those are company-reported scale claims. The site also says it is GDPR and SOC 2 compliant and EU hosted; a buyer should verify the applicable scope and controls through diligence.
Why I’d look closer
The founder background matches the product. YC identifies James He as a computational social scientist who led a large study of LLM societies and previously built lending ML systems at Yonder. Patrick Sharpe is described as an applied behavioural scientist who worked on enterprise research and customer-experience projects at iptiQ and Unconventional Wisdom.
The advantage is speed and scenario breadth. A comms or strategy team can test many messages, audiences and second-order reactions before commissioning a full human study. The tradeoff is model validity: a simulated regulator, investor or customer is still a model. It can surface hypotheses, language risks and comparisons, but it should not be treated as a vote, a forecast or proof that a real stakeholder will behave the same way.
What I’d ask
Which real-world data grounds the personas and how often is it refreshed? Can the team separate generated signal from observed survey or behavioural data? What is the confidence interval around a segment difference? How are sensitive strategies isolated, retained and deleted? Can an analyst reproduce the same scenario and see which assumptions drove the result?
My editorial take
Shortlist Artificial Societies when the decision is expensive, the audience is hard to recruit and you need a fast map of plausible reactions before doing human validation. Use the output to sharpen questions, messages and test cells—not to skip the people whose behaviour determines the outcome. The product is most credible as a scenario engine with an evidence boundary, not a synthetic replacement for the market.
Quick facts
| Field | Sourced detail |
|---|---|
| Product | AI audience simulations with surveys, interviews, conjoint and scenario analysis |
| Buyer | Enterprise marketing, comms, strategy, research and consulting teams |
| Use cases named | Reputation, crisis communications, product positioning and investor communications |
| Security claims | Site says GDPR/SOC 2 compliant and EU hosted; scope requires verification |
| Main question | What decision will the simulation improve before real-world validation? |
Sources checked
| Source | Checked |
|---|---|
| YC company profile | 2026-09-19 |
| Artificial Societies homepage | 2026-09-19 |
| Case studies | 2026-09-19 |
| Artificial Societies YC launch | 2026-09-19 |
Cohort context
Artificial Societies is listed in Winter 2025. In our 2026-09-18 directory snapshot, 104 of 165 listed companies in that cohort have YC’s primary industry label B2B (63.0%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:19:23.370Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Observed |
| Docs/developer link | Not observed in this response |
| Pricing link | Not observed in this response |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗