Daveed Gartenstein-Ross is an author, professor and counterterrorism expert who began experimenting with simulation games while teaching at Georgetown University in 2013. He found that putting students into interactive scenarios helped them understand complex geopolitical situations in ways traditional classroom instruction couldn’t.
“The human brain has trouble understanding [situations in their full depth] from a classroom discussion or from reading,” Gartenstein-Ross said. “It’s often experiencing it that gives you that answer.”
Gartenstein-Ross carried that approach into his previous venture, Valens Global, where simulation games became popular with customers but were labor-intensive, requiring considerable time to research, build and perfect.
“Our games were always hailed as the best thing that our customers had done for training or for education, but we were always losing money on them,” Gartenstein-Ross said.
Before exiting Valens Global in 2024, he realized advances in AI could dramatically reduce that workload. That realization led him to spin off the gaming practice and launch ExpertTheory.
This new startup, officially launched in 2024, is now using AI to transform how educational and professional simulation games are created, turning a process that once took weeks or months of intensive work into one that can take just a few hours—or as little as five minutes, for a basic game.
Game creation and structure
ExpertTheory’s games divide participants into teams representing different stakeholders in a simulated scenario, each with their own objectives, capabilities and information. Teams make decisions and communicate with one another over a series of turns, with the platform helping determine the consequences of their actions and how the scenario develops.
The process of creating a game begins with a three-part questionnaire that asks users to establish the simulation’s time period, background, central focus and participating teams. Gartenstein-Ross said the questionnaire, which currently uses static prompts, will soon get a “facelift” with a more dynamic, conversational AI interface.
From there, the company’s AI-powered platform, which uses proprietary code to guide game creation and management while drawing on an external large language model to generate content, produces a playable game using both the information provided and the model’s broader knowledge.
While a basic game can be generated in under five minutes, Gartenstein-Ross said building a more complex simulation typically takes around 90 minutes.
“AI is very smart, but… an informed expert is going to be sharper than AI alone,” Gartenstein-Ross said, emphasizing the company’s focus on human-AI collaboration.
The platform can also incorporate specialized information through retrieval-augmented generation (RAG), enabling users to prioritize particular documents when building a simulation. Rather than limiting the AI to those sources, the system draws on its broader knowledge to flesh out the game. Indeed, given the speed at which ExpertTheory can create simulations, they can even do things such as turn a piece of literature into a game.
Notably, ExpertTheory has already been commissioned by think tank New America to develop a game based on counterterrorism expert Peter Bergen’s book All the President’s Wars.
“That’s kind of an example of where we’re taking specialized knowledge and also pulling from the world and creating a game based on that,” Gartenstein-Ross said.
Current projects
As of this writing, ExpertTheory is at work on a major project at Duke University, based on a game originally designed by Eleanor Ross, the startup’s co-founder and creative director. Dubbed Arctic Awakenings, the game is being used in Professor David Schanzer’s full-semester course, with Gartenstein-Ross and Alice de Saint-Aignan serving as referees.
The experience places students in a simulated geopolitical competition set in 2030, when climate change has made the Arctic navigable year-round. Students are divided into five teams, each receiving background materials, confidential memos and updates through a simulated news and social media environment.
Over five turns, teams make strategic decisions, communicate with one another and respond to developments, while human referees use ExpertTheory’s AI-powered platform to evaluate their actions and advance the scenario.
QUICK BITS
Startup: ExpertTheory
Co-Founders: Daveed Gartenstein-Ross, Eleanor Ross
Founded: 2024
Team size: 13
Location: Raleigh
Website: experttheory.com
Funding: Closed SAFE round, raising pre-seed
ExpertTheory is also running a capstone exercise for Arizona State University’s master’s program in global security and is working with approximately 35 universities overall.
Beyond higher education, ExpertTheory has worked with the U.S. Air Force, with which it recently secured a contract worth more than $200,000; National Defense University and the Naval Postgraduate School; and private sector companies such as Google’s Jigsaw.
(Editor’s Note: These projects were among the factors that made ExpertTheory a judges’ selection for GrepBeat’s own 2026 Startups to Watch Awards.)
Expansions
ExpertTheory is exploring a marketplace model, working with users who have created games on its platform to commercialize them through joint ventures. Gartenstein-Ross hopes this will eventually create a diverse catalog of games spanning different industries and subjects.
Looking forward, Gartenstein-Ross sees the company’s biggest growth opportunity in fully AI-facilitated games, which would allow users to play without needing a human referee to manage the simulation.
The company is preparing to test its first fully AI-facilitated game and expects to begin selling standalone games by the end of October. Rather than requiring customers to subscribe to the platform and create their own simulations, Expert Theory would be able to sell individual games that are ready to play.
“I think that sale of games and not just a platform to create games is probably the true north,” Gartenstein-Ross said.
