For decades, if you wanted to know whether consumers would buy your new shampoo or laugh at your ad campaign, you rounded up a dozen people in a windowless conference room and asked them. Then you waited. And waited—for transcripts, for analysis, for the invoice.
Foresight, a startup that just graduated from Y Combinator's Spring 2026 batch, thinks that entire ritual is headed for obsolescence.
The New York startup has built software that simulates consumer panels using what it calls "digital twins"—synthetic personas constructed from demographic and behavioral data. Query the system about packaging designs or pricing strategies, and the company claims it delivers predictions in minutes. No recruitment. No scheduling conflicts. No stale bagels.
Foresight claims 95% accuracy when benchmarked against actual human responses, though this figure is self-reported without independent third-party validation—more on that in a moment.
Caudalie, the French beauty brand, has apparently gone all-in—at least according to a case study published on Foresight's own website. The company now runs roughly 15 tests weekly on Foresight's platform and says it's ditched traditional focus groups entirely. Alix Ligot, Caudalie's Global Marketing Development Director, is quoted praising the platform's ability to deliver insights in under 10 minutes at one-twenty-fifth the cost of conventional methods.
That's the pitch, anyway. Whether synthetic consumers can truly mimic the messy, contradictory reality of actual shoppers—well, that's where things get interesting.
Personas Built From Census Data
Foresight's approach starts with numbers. The platform assembles synthetic consumer personas using census and socioeconomic data from 47 countries, pulling from sources like the U.S. Census Bureau, Eurostat, and Brazil's IBGE. Those personas are then weighted using iterative proportional fitting—a statistical technique also used by Gallup, Pew, and Ipsos to balance survey samples.
Founders Eytan Rozenblum and Antoine Bertrand say they've tested their system against more than 18,000 real respondents across over 20 studies. In one blind test with a Fortune 500 company—involving more than 100 paired estimates—Foresight claims it hit a Lin's concordance correlation coefficient of 0.88, a statistical marker the company argues is high enough to replace traditional fieldwork.
The use cases are standard CPG fare: concept testing, packaging evaluation, price elasticity studies, brand lift measurement, campaign tracking. The kinds of questions brands ask constantly but often can't justify the expense or timeline of formal research to answer.
Fixed Pricing in a Custom-Quote World
Here's where Foresight diverges from the industry norm. Traditional market research typically involves bespoke quotes and agency relationships—call it opaque pricing at best. Foresight offers monthly subscriptions with fixed tiers (as of mid-2026, subject to change).
Entry level: $479 monthly, 10,000 credits, 25 pre-built audience segments. Growth: $879 monthly, 100,000 credits, 50 segments. Enterprise customers get custom segmentation, single sign-on, and what the company calls "white-glove serviced studies."
Both self-serve plans include seven-day free trials. When Foresight launched on Y Combinator's platform in late May, it dangled the first month free on any three-month pilot signed before May 30.
The company displays SOC 2 Type II and GDPR compliance badges on its homepage. No external audit documentation was publicly available at publication.
What the Platform Can't Do
To their credit, Rozenblum and Bertrand acknowledge limitations upfront. The platform can't handle product tests requiring physical interaction—no taste tests, no texture evaluation, no fragrance trials. It also warns against using synthetic panels for legal compliance work, safety-critical UX testing, or gauging reactions to radically novel concepts with zero consumer reference points.
Then there's the bigger caveat: Foresight's accuracy claims are self-reported. No independent third-party audits. No peer-reviewed validation of the 95% accuracy figure or the Lin's CCC metric. The Caudalie case study, while detailed, lives on Foresight's own website with no external confirmation from the beauty brand's PR team.
This isn't exactly unusual for young B2B software companies. But it does mean buyers are essentially taking the founders at their word until they run their own pilots.
Everyone Wants a Piece of This
Foresight has plenty of company in the race to replace traditional consumer research. BluePill announced a $6 million seed round last November to build similar "AI consumer twins," claiming 90% cost reduction. Toluna launched HarmonAIze Personas in February 2025. YC's Winter 2025 batch included Artificial Societies, building multi-agent simulations of target audiences.
The incumbents aren't sleeping either. NIQ rolled out its BASES AI Screener in April 2025 to accelerate concept testing. First Insight launched "Ellis" in January 2026, promising to compress weeks of insights work into minutes. Circana introduced its "Complete Why" AI-enabled suite in March.
The momentum is real. A March 2026 NIQ report found that 54% of CPG companies now use AI for research purposes. Deloitte's 2026 Global CPG Outlook flagged the industry's pivot toward AI-powered approaches across marketing and product development. This technology has crossed over from experimental to expected.
Speed as Strategy
What sets Foresight apart—if anything does—seems to be pure execution speed. Traditional research might take weeks from kickoff to deliverable. Foresight promises answers before lunch. Focus group costs can climb into five figures for a single study; Foresight offers unlimited queries within a monthly subscription.
Can synthetic consumers truly replicate the irrational, unpredictable, often self-contradictory behavior of real people? Open question. But for brands running dozens of small tests each week—decisions that don't warrant $50,000 studies but still benefit from consumer signal—the economics might be persuasive enough.
Rozenblum and Bertrand are wagering that speed and cost will outweigh perfect fidelity. If Caudalie's reported usage is accurate, at least some CPG teams are ready to make that bet.
The focus group isn't dead yet. But it might want to start updating its résumé.
