Skip to content

Aaru

4 viewers · 30d

Total raised

$88.0M

2 filings since 2024 · latest Equity filed

Cumulative raised
LAST ROUND
Equity · $80.0M
ROUNDS
2
INVESTORS
22
FOUNDED
2024
HQ
New York, NY
SECTOR
AI
EMPLOYEES
N/A
30D VIEWERS
4

AI overview

Updated

Aaru builds large-scale behavior simulations that predict how real populations will respond to a product, price, message or strategy before a company commits to it. The platform generates populations of AI agents - roughly 10,000 per audience - grounded in public datasets such as census and labor statistics, licensed private behavioral data such as transaction and search patterns, and the customer's own context, then runs hypothetical scenarios against them. It is positioned as a replacement for surveys and focus groups, on the argument that simulated behavior predicts action better than self-reported preference. Founded in March 2024 by Cameron Fink, Ned Koh and John Kessler, the company operates from New York and Singapore and counts Accenture, EY and Interpublic Group among its customers.

What sets it apart

Populations are trained simultaneously at the individual and aggregate level so trait relationships are preserved rather than sampled independently, and results are validated against real-world outcomes rather than against survey answers - an EY engagement reproduced six months of global wealth research in a day at 0.90 median correlation, and the company's agent-based polling called the 2024 New York Democratic primary.

Funding history

2 rounds
EquityDec 17, 2025
Form D
+$80.0M$88.0M total
SAFE+1Apr 3, 2024
Form D
+$8.0M$8.0M total

Products

1 tracked

Aaru Simulation

AI market research platform

A simulation platform that builds synthetic populations from real demographic, behavioral and outcomes data and runs hypothetical scenarios against them, returning interactive analysis of how the population would respond.

  • Guided workflow from objective and audience definition through population build, simulation run and interactive analysis
  • Populations of about 10,000 agents per defined audience
  • Multi-source grounding: public datasets, licensed behavioral data, and customer context
  • Trait-relationship preservation, with joint training at population and individual level