×
Simile
Enter password to continue
×
Simile
Loading…
×
Simile
Stripes is an NYC-based growth equity firm led by entrepreneurs and operators. We invest in and actively support best-in-class, category-defining companies. We believe Simile will redefine how the world understands consumers.
Stripes partners with founders building n-of-1, category-defining companies with enormous opportunity. We look for amazing products with amazing market opportunities.
| Company | Overview | Round |
|---|---|---|
| Leading data infrastructure platform | $10B, Dec 2024 | |
| Finance automation platform | $150M, Dec 2024 | |
| Autonomous mobility solutions | Undisclosed | |
| AI-powered public safety | $275M, Mar 2025 | |
| Premium athleisure brand | $825M, Nov 2024 |
Consumer research runs on the survey panel. That foundation is eroding. Response rates have collapsed, fraud has scaled with automation, and professional respondents now flood the sample sources brands pay for.
| Failure | What the data shows | Source |
|---|---|---|
| Response-rate decline | Pew phone response rates fell from 37% (1996) to 9% (2014), then to 6% (2018) | Pew Research Center |
| Usable data collapse | Open online surveys once expected >75% usable responses; today only 10% of studies can; usability now lingers in the 0–25% quartile | Frontiers in Research Metrics and Analytics |
| Fraud distorts results | Fraud shifted Marriott aided awareness from 71% (qualified respondents) to ~59% (flagged respondents) | GreenBook / Rep Data |
Survey fraud was projected to cost the industry roughly $350M in incentive spending in 2024, about 5% of total reward spend, per ResearchShield citing Tremendous — directionally consistent with the first-party fraud data above. Sources: Pew Research Center; Frontiers in Research Metrics and Analytics (Pinzón et al., 2024); Rep Data "State of Fraud 2025" via GreenBook.
Even when the panel works, the economics discourage using it. A single concept test costs tens of thousands of dollars and takes weeks. Brands ration research to the few decisions that can absorb the cost and the wait.
| Study type | Cost | Timeline | Source |
|---|---|---|---|
| Concept test (~250 respondents) | $43K–$79K | 2–3 weeks | MX8 Labs |
| Quarterly brand tracker (400/wave) | $150K–$240K/yr | Ongoing | MX8 Labs |
| Single online survey (400–1,000 completes) | $5K–$15K | 1–2 weeks | MainBrain Research |
| Qualitative (15–20 IDIs) | $40K–$65K | 3–5 weeks | MX8 Labs |
Cost figures span vendor and agency sources; ranges vary with market count and scope. MX8 Labs is an AI-research vendor publishing its own cost analysis; MainBrain Research is a traditional agency. Sources: MX8 Labs "What Market Research Actually Costs"; MainBrain Research "How Much Does Market Research Cost in 2025."
Research teams held out longer than most functions. That has changed. Adoption is now near-universal, and researchers expect synthetic responses to become the default within a few years.
| Finding | Detail | Source |
|---|---|---|
| Investment intent | 83% say their organizations plan to significantly increase AI investment in 2025 | Qualtrics |
| Quality remains the gate | 40% of researchers rank data quality as their top challenge — the barrier synthetic data must clear | GreenBook GRIT |
| Vendor speed claims | Synthetic-research vendors claim timelines from "6 months to 6 hours" and 88% average similarity vs traditional research — self-reported and unaudited | Evidenza (vendor) |
Vendor accuracy claims are self-reported; no industry-standard synthetic-research accuracy benchmark exists. Sources: Qualtrics "2025 Market Research Trends Report"; GreenBook "The Sea Change: 2025 GRIT Business Outlook"; Evidenza.ai.
A broken panel and an eager buyer are not enough on their own. What makes this an investable moment is that the underlying technology crossed a threshold. Three shifts now compound.
Simile is the direct commercialization of the Stanford generative-agents research program. The academic work moved from simulating communities to simulating individuals, and the accuracy is now high enough to inform real decisions.
The current model is episodic. A brand scopes a study, waits weeks, and receives a deck. Simulation changes the unit of consumption from a project to a query.
Fully synthetic populations are fast but ungrounded. The approach that holds up is one that anchors simulated agents in data from real people.
Consumer research today is where drug discovery was five years ago. The tooling changed before the industry did. The best teams have new instruments; the incumbent workflow has not caught up. That gap is the opportunity.
Source: Research World (ESOMAR), "Inside the $153bn Insights Industry," 2025.
We spoke with senior insights and marketing leaders across the Stripes network. The pattern was consistent: the current model is too slow and too expensive, panel quality is getting worse, and buyers are open to AI but will not adopt it without proof.
Simile's tagline is direct: "Simile is a simulation platform for human behavior." We believe the team, the method, and the timing line up to make that platform the category standard.
Simile is built by the team that created the field. Its founding work introduced generative agents and the term "foundation model," and its simulations are validated against held-out human data.
Simile partners with real people to build high-fidelity models of how each of them lives and makes decisions, then orchestrates those digital twins to answer what real people will do and why.
Panel quality is collapsing at the same moment insights teams have adopted AI and expect synthetic responses to become standard. Both curves point the same way at the same time.
The founding team is the source of the underlying science, and the round reflects it: a $100M Series A led by Index Ventures, announced February 12, 2026.
The wedge is the insights team, but the surface is the whole consumer economy — every decision that today ships without a test.
Three models compete for the same budget: grounded simulation, pure synthetic prediction, and the legacy human panel. They differ on the one thing buyers care about — whether the answer holds up.
| Dimension | Simile (grounded simulation) | Pure-synthetic shops | Legacy panels |
|---|---|---|---|
| Grounding | Digital twins built from real people, anchored in Gallup's nationally representative panel | Statistically generated agents, no per-person grounding | Real humans, but a shrinking and fraud-prone sample |
| Scientific basis | Founding team created the generative-agents field and coined "foundation model" | Applied ML, no primary research lineage | Established survey methodology |
| Speed | Query a standing model; answers in the time to ask | Fast | Weeks per study |
| Validity story | Validated against held-out human responses; fidelity is the product | Vendor-reported accuracy, unaudited | Ground truth in principle, degrading in practice |
| Framing | "A simulation platform for human behavior … a foundation model that predicts human behavior in any situation" | Replace surveys with prediction | Sell access to respondents |
Positioning language quoted from simile.com and Index Ventures. Source: simile.com; indexventures.com.
Synthetic and simulated research has drawn a wave of startups and a fast response from the panel incumbents. The field splits into fully synthetic prediction, grounded twins, and AI-accelerated human data collection. Simile sits with the grounded twins.
Aaru simulates statistically representative populations trained on behavioral outcomes rather than survey self-reports, replacing surveys and focus groups with predictive modeling. It is the clearest contrast to Simile's grounded approach.
Sorted by disclosed funding. Badges mark relative standing.
| Company | Positioning | Differentiation | Funding |
|---|---|---|---|
| Brox.AI Contender | Persistent panel of 60,000 digital twins of real people, surveyable instantly and repeatedly | Closest structural analog to Simile's real-person grounding, but panel-as-product not platform | Undisclosed; after ~10x revenue growth |
| Artificial Societies Emerging | Simulates whole social networks to test how messages propagate (YC W25) | Society-level interaction effects vs individual-respondent simulation | ~$5.35M pre-seed + seed |
| BluePill Emerging | "AI consumers" from behavioral signals, testing products, packaging, and creative | CPG and creative-testing focus, behavioral-signal twins | $6M seed (Nov 2025) |
| Blok Emerging | Synthetic users that interact with prototypes and live products | Behavioral product-interaction, not attitudinal research | $7.5M seed (Jul 2025) |
| Evidenza Contender | B2B "silicon samples" plus synthetic-expert advisors; ex-LinkedIn B2B Institute founders | Only notable B2B-focused player, where human sample is hardest to recruit | Bootstrapped, profitable |
| CulturePulse Emerging | Psychologically realistic digital twins; quantifies emotion and personality across 80+ languages | Cognitive/psychological modeling heritage; named an Aaru competitor | $1M seed + ~€1.5M |
| Yabble Exited | "Virtual Audiences" synthetic respondents; early mover | Acquired by YouGov for £4.5M — panel incumbents are buying, not building | NZ$3M, then acquired |
| Synthetic Users Emerging | Synthetic participants for interviews, surveys, and usability studies | UX-research entry point for product/design teams | Undisclosed |
| Listen Labs / Outset / Keplar Adjacent | AI-moderated interviews of real humans at scale | Human-in-the-loop data collection vs simulated respondents | Varies |
The panel and insights incumbents are moving into synthetic data — validating the category and defining the distribution Simile must beat.
| Incumbent | Move | Stance |
|---|---|---|
| Qualtrics | Launched synthetic consumer panels at X4 (Mar 2026): an LLM trained on 200M+ research respondents, claiming 12x better accuracy than general-purpose LLMs at ~half the cost of human panels | Leader |
| Kantar | Synthetic data a core focus of its AI Lab; ships "synthetic data boosting" for brand tracking claiming 94–95% accuracy vs ground truth, building twins for Brand Guidance | Leader |
| NIQ / Nielsen | BASES AI (Sep 2024): synthetic respondents on consumer-permissioned household data plus a CPG LLM; a 2026 Reckitt case study reported up to 65% faster research | Contender |
| Ipsos | Partnered (Aug 2025) with Stanford's Robb Willer — a co-author on the 1,000-person paper — for validated digital-twin panels; notes twins show lower emotional depth than real respondents | Contender |
| YouGov | Acquired synthetic-response pioneer Yabble for £4.5M — the clearest buy-vs-build move among panel incumbents | Contender |
| SurveyMonkey | No synthetic product; positioned against the trend. Nearly doubled its human panel to 335M+ and published research showing 51% of people distrust synthetic responses | Skeptic |
The entry point is the insights team. From there the product expands into every decision that today ships untested, and across the verticals where Stripes already operates.
From a single team to a standing capability across the organization
Land with concept and message testing
Always-on read replaces quarterly waves
Concept and product pipelines
Pre-test every asset and placement
Test price and range before launch
The first expansion is turning episodic studies into a standing query surface — brand tracking that updates continuously instead of quarterly.
When the cost per question falls, testing spreads from a few big bets to routine practice across creative, media, packaging, and pricing.
The consumer economy is the addressable surface: CPG, beauty, fashion, restaurants, retail, and media.
Stripes' in-house Scale Team provides hands-on support across every critical function — capital plus execution embedded alongside your team.
Executive search, org design, compensation benchmarking, employer branding
Brand positioning, content strategy, demand generation, event marketing
FP&A setup, audit readiness, legal counsel, entity structuring, tax strategy
Data infrastructure, KPI dashboards, customer analytics, pricing optimization
GTM strategy, enterprise sales playbooks, pipeline management, partnership sourcing
Procurement, vendor management, international expansion, systems & process design
Design and staff the enterprise sales motion for insights and marketing buyers — territory planning, pricing strategy, and customer success frameworks
Source senior commercial leaders with research and enterprise-software experience — CRO, VP Sales, Head of Partnerships — through our network
Open doors to insights and brand leaders across our consumer portfolio — warm intros to the exact buyers Simile is built for
Establish Simile as the category standard in AI-based simulation — conference strategy, media relations, case-study development, and analyst engagement
Stripes has spent two decades backing consumer brands and, more recently, the AI infrastructure they run on. The companies we know best are exactly the companies Simile is built to serve.
Fashion, restaurants, CPG, beauty, media, retail — every one a consumer-research buyer
Also BRUNT, Jacques Marie Mage, The Black Tux, Popup Bagels, 7th Street Burger, Fishbowl, La La Land Kind Cafe, Snooze, Seven Sundays, Stella & Chewy's, Harvest Hosts, Shibumi Shade, Aviron.
Direct pattern recognition for backing an AI-native platform
Simile sells to insights and marketing leaders at consumer companies. Our network reaches them directly — portfolio operators, exited founders, and executive advisors across the consumer economy.
Warm intro through a portfolio food-and-beverage brand's insights leader, who runs exactly the trackers and concept tests Simile replaces
Direct path via a founder in our beauty and apparel portfolio who owns brand and consumer research decisions
Introduction through our Tech Council to the CIOs and CDOs who control AI procurement at large consumer enterprises
×
Simile
We believe Simile will redefine how the world understands consumers. Let's build it together.