SAPIENS 2.0 release
Next-generation model, trained on the widened source graph.
33% better than standard LLMs and other behavior foundation modelsProduct updates from the Vectorial AI team. Every capability shipped on the SAPIENS platform since day one, across three fronts: the behavior model itself, the simulation suite on top of it, and the data sources underneath.
Next-generation model, trained on the widened source graph.
33% better than standard LLMs and other behavior foundation modelsFast, lower-cost tier opening simulation to lighter-weight use cases.
AI-run interviews with real participants, feeding enriched signal back into the model.
Consumer intelligence partnership deepening India-market coverage.
Formalized inbound data partnerships to compound the data moat.
Real-human panels wired in to ground, benchmark and calibrate SAPIENS.
Always-on segment insights layered on top of the simulation engine.
Variant testing across segments, pick the winner before the traffic split.
Simulated user response on Lovable builds, before they reach real traffic.
Simulation run directly on designs, where design already happens.
Conversion and messaging response predicted per audience segment.
Product-led prototype testing, simulate how users react before you build.
Simulated chat experiences: how users converse with a product, predicted.
Ad and creative concept testing against simulated audience segments.
Populations self-organize into behavioral tribes, segmenting the audience automatically.
Full production model: higher behavioral fidelity, larger source graph.
25% better than standard LLMs and other behavior foundation modelsReal-time social discourse added as a grounding source for personas.
Expanded the behavioral corpus with two high-signal public platforms.
Customers' first-party data connected directly into the simulation loop.
Predicted adoption and reaction testing for product concepts pre-launch.
Step-by-step behavioral walkthroughs across the full customer journey.
Synthetic survey fielding with response distributions in hours, not weeks.
Multi-persona qualitative sessions, moderated end-to-end by SAPIENS.
Professional and community discourse added as panel-sourcing platforms.
Agentic research that assembles real user panels from across social media.
First release of our proprietary human-behavior prediction model.