Vectorial raises $3M to simulate users before launch

Vectorial raised $3 million in seed funding to make behavior simulation a step before building, the company said October 20.

The round was backed by Smilegate and SenseAI Ventures alongside Berkeley SkyDeck, Berkeley AI Research Lab, the Stanford GSB fund and Dent Capital, with product leaders from Meta, Shopify and Mastercard participating. As of October 2026 the company tied its academic roots to Stanford StartX, CMU Venture Bridge and UC Berkeley.

That pitch is speed. Vectorial said teams can test a flow, campaign or agent against thousands of modeled users in minutes.

Taranveer Singh, co-founder and chief executive, framed the problem as two AI frontiers and said his bet is on the second. He said reasoning gets attention while simulation and decoding human behavior is the other gap, and he argued the missing pieces are the ability to simulate many scenarios and to bring users into the build room.

The mechanism is SAPIENS. Vectorial described it as a behavior model that learns latent traits across socio-demographic, psychographic and behavioral dimensions for a target customer, then generates modeled users grounded in real behavior data.

The company said SAPIENS is 30 percent more accurate in predicting behavior than general-purpose LLMs, outperforms academic models Be.FM and Centaur, and is validated against real-user studies. Vectorial said it has modeled more than 2.5 million real people and run more than 30,000 simulations to date.

Jasmine Kaur, co-founder and chief business officer, said enterprises treat a continuously enriched audience model as a strategic investment. She said that is why adoption spreads across product, design and marketing, because the teams design for the same audience.

Rahul Garg, co-founder and chief product officer, pushed back on the idea that a better model alone is enough. He said Vectorial puts behavior models to work inside the development loop and that this changes how products are designed and built.

Singh contrasted that approach with the two defaults he sees now. He said AI-moderated interviews still take days for 50 to 100 participants and every new question restarts the clock, while synthetic agents answer instantly but lack grounding in real behavior data. Other simulation firms focus on market research for strategy, he said, not on downstream product decisions.

Customers so far include MIDI Health, GoodRx, Suki AI, Boat and Concrete Software, per the company. Vectorial said those teams run around 600 simulations a month, equivalent to about 40 user studies a month and 20 times more research than before, and that simulations have reproduced prior study findings with more than 85 percent accuracy.

Supreet Pal Singh, chief technology officer at Everkind, said his team holds a high bar for behavior and alignment but a conversation that works for one person can land differently for a first responder or a tired parent, and Vectorial lets them test those differences end to end before an update reaches real users.

Vedansh Kumar, chief marketing officer at Boat, which the company described as $350 million in revenue and valued at $2 billion in India, said consumer research has historically informed only a fraction of brand decisions and the promise here is to bring that understanding into everyday choices.

David Nixon, vice president of Data Services at Concrete Software, maker of PGA Tour and NHRA Drag Racing games, said the value is knowing what to build first and getting a first read in minutes instead of days or weeks.

The gap between those testimonials is how much trust a simulation can carry. Vectorial leans on scale and grounding in real data, Singh leans on speed, Garg leans on workflow fit. None of them detailed in the announcement how consent, privacy or refresh cadence works for the 2.5 million profiles, or how SAPIENS stays current with what the company called new emerging behavior.

Modeling human behavior has a long baseline outside software. Research published on Tinshemet Cave found Neanderthals and Homo sapiens shared technology, lifestyles and burial customs as early as 110,000 years ago in the Levant, according to ScienceDaily, a reminder that interaction patterns leave traces long before a product funnel exists.

Vectorial said it will use the funds to expand behavioral datasets through new data partnerships and to build SAPIENS 2.0 as simulation infrastructure. Rahul Agarwalla, managing partner at SenseAI Ventures, said building has moved to lightspeed while understanding users has not, and he described Vectorial as replacing surveys with millions of simulated humans.

The question the announcement did not answer is what happens when a simulation is wrong at scale.