Tell us about your company and the problem it solves, or its benefit to learners or educators.
CoChat is an AI research workspace built for higher education, on a mission to solve the trust problem that has followed AI into academia. These tools have made research faster, but they also fabricate sources, inventing papers, generating fake DOIs, and misattributing quotes. Faculty have been left to choose between banning AI outright or fact-checking every reference by hand, while students learn research habits on systems that regularly mislead them. CoChat verifies every citation, quote, and paraphrase against CrossRef and Semantic Scholar before it reaches a draft, drawing on a database of more than 200 million papers. In independent benchmark testing, CoChat also scored significantly higher on citation accuracy than the same base AI model working without verification, which shows how much of the problem lives in the workflow around the model rather than the model itself. Researchers ultimately get the speed of AI without putting their credibility at risk.
Where did the idea come from to create your company?
CoChat began as an AI collaboration platform for teams, a shared workspace where colleagues could work with AI and develop agents together instead of in isolation. As the platform grew, we noticed graduate students, PhD candidates, and educators showing up in our user base in meaningful numbers.
When we talked to them, the same frustration surfaced in almost every conversation. They valued what AI could do for literature review and analysis, but they could not trust the citations it produced, and in academic work a fabricated reference can do real damage to a career. 1 in 4 citations in literature are wrong and AIs confidence hides this problem in plain sight. I came to understand that researchers needed a tool that they could verify, which reshaped the company and led us to sharpen the product capabilities for academic rigor.
Tell us about one challenge and how you overcame that challenge.
The hardest challenge was resisting the industry’s default playbook. Nearly every AI product competes on speed while targeting business use cases, and we chose to compete on verification, which is slower and much harder to build. Checking every quote, paraphrase, and DOI against scholarly databases in real time is a genuine engineering problem, and for a stretch it felt like we were swimming against everything the market claimed to want. The benchmark work carried us through. When independent testing showed our verified workspace beating the same underlying AI model by a wide margin on citation accuracy, we had proof that rigor and usability could coexist in one product. That evidence gave us the conviction to keep building for trust first, and the response from the academic market confirmed we had read the moment correctly.
What are you most proud of or what is your company’s greatest achievement?
I am most proud of how organically the platform has grown to date. Students, educators, and researchers at more than 100 universities, including Michigan, Cambridge, Columbia, NYU, and Florida State, found CoChat and began using it without any institutional mandate behind them. Tools usually arrive in higher education from the top down through procurement committees and pilot programs. Ours has spread from the bottom up because individual researchers decided it made their work more trustworthy, and that is the hardest audience in the world to win over. When a PhD candidate stakes their dissertation citations on your platform, that carries more weight than any contract or press release ever could.
Where do you see your company in five years?
I believe citation verification will be as standard in five years as spell check is today, and I want CoChat to be the reason that happens. More broadly, we see the entire research workflow consolidating, with discovery, literature monitoring, analysis, verification, and writing coming together in one trusted workspace instead of a patchwork of disconnected tools.
Our goal is to become the platform where the next generation of scholars learns to do rigorous work with AI, and to help the field develop the shared reliability standards it still lacks. Higher education is moving past the question of whether to allow AI and into the harder question of how to govern it well. In five years, I want CoChat to be the infrastructure that made that transition possible.