Harmonic just launched a beta chatbot app on iOS and Android. The app gives users access to its AI model, Aristotle, which claims to deliver “hallucination-free” answers in complex math reasoning.
Harmonic co-founder and CEO Tudor Achim says Aristotle is designed for “mathematical superintelligence” (MSI). It aims to help users with math-heavy fields like physics, stats, and computer science.
Harmonic backs its bold claim with a built-in verification process. Aristotle outputs answers in the open-source Lean programming language, then double-checks them with a non-AI algorithm. This formal verification approach is similar to what’s used in aviation and medical device software. 
Harmonic says this method guarantees no hallucinations in quantitative reasoning tasks — a tough feat given that even top AI models today struggle with hallucinations, and the problem isn’t improving. Achim emphasized this in an interview with TechCrunch:
“[Aristotle] is the first product available to people that does reasoning and formally verifies the output,” said Harmonic CEO and co-founder Tudor Achim in an interview with TechCrunch. “Within the domains that Aristotle supports, which are quantitative reasoning domains, we actually do guarantee that there’s no hallucinations.”
The startup just raised $100 million in a Series B led by Kleiner Perkins, valuing it at $875 million. Achim says Harmonic is moving fast toward MSI and investors see the potential.
Harmonic also plans to launch an API for enterprises and a web app for consumers.
Aristotle reportedly scored a gold medal on the 2025 International Math Olympiad in a formal, machine-readable test. Google and OpenAI’s IMO gold medal models took informal, natural language tests this year.
The launch comes amid intense competition to build reliable AI reasoning tools — with no clear leaders yet.
The beta rollout makes Aristotle available to more people now — with a focus on real math and rock-solid answers.
TechCrunch event
San Francisco | October 27-29, 2025
