Your methodology. Your client. Our engine, white-labeled.
Agencies are being asked for AI capability, but client data-science teams ask for validation. Building the engine in-house takes time and expertise. Licensing a vendor with marketing claims but no papers creates review risk. Simulacra is the third option: a validated engine, headless via API, with the validation papers attached.
Build vs. buy? Both options are bad.
Building takes too long
Two AI engineers, four quarters, and a methodology team to ship something already behind. Your competitors won't wait.
Most vendors won't survive review
Synthetic-respondent vendors are winning on the demo and failing on the client's validation. Don't let your agency own that failure.
Client expects a client-ready output
Cross-tabs alone are not enough. Methodology appendix, validation diagnostics, and assumption documentation have to ship with every deliverable.
API, docs, validation papers, and white-label delivery.
Headless API integration
OpenAPI, Auth0 M2M, idempotency keys, async jobs, stable errors, and schema-first conditioning for agency data pipelines and workflows.
Methodology appendix on every output
Holdout backtest results, performance by subgroup, known limits, recommended use. Co-brandable; client-ready.
Validation papers, public
Twin-2K-500. Pricing & Promo. Data Reduction. Show your client the same papers we showed you. The math is in the open.
Audit-grade governance
Synthetic-row labels. Scenario logs. Tenant-scoped data. SOC 2 / ISO 27001 documentation for procurement.
Standard exports
SPSS, Q, Tableau, Power BI, R, Python — the formats your reporting workflow already consumes.
Commercial flexibility
Usage-based pricing. Co-marketing where it serves both sides.
Three published validations: the hardest public benchmark, data reduction, and pricing & promo predictive forecast, with more in progress.
Twin-2K-500
Hardest public benchmark in synthetic survey data. Beaten below sampling noise.
Applied benchmarkPricing & Promo
Real Fortune-class CPG. Blind, asynchronous. Customer signed off.
Economic benchmarkData Reduction
80% reduction in sample requirements. Bayesian changepoint corroborated.
Agencies need unreasonably effective AI for unreasonably demanding clients.
Build a client workflow on the API — or let an AI agent run it for you, end-to-end. Computer to computer, no human handoff required.