Theorem blog · Feb 2026
Describes Theorem’s verified translation benchmark, rocq-dove specification generation, and a 350x speedup in verified software translation.
Program verification so even your systems engineers can vibecode
Public company, workplace, funding, and market signals
Updated Jul 29, 2026
Theorem is a San Francisco-based AI and programming-languages research lab building trustworthy-by-default software tools that use formal verification and AI to prove code correctness, especially for mission-critical and AI-generated software.
Primary product
AI-powered program verification and proof-generation tools
Founded
2025
Headquarters
San Francisco, California, United States
Team size
1-10 employees
Industry
Software Development
Sub-industry
AI software verification / formal methods
Offices
0 jobs at Theorem
Check back later for new openings
Business model
Stage
Seed
Total raised
$6M
Latest round
Seed · Jan 2026
Latest amount
$6M
Jan 2026 · Khosla Ventures
Investors
Small, fast-growing team of engineers and researchers focused on unusually hard, high-impact problems and early adoption of new techniques.
Compensation
Public job posts cite roughly $150K-$250K for ML research engineer roles and $300K-$500K for ML research scientist roles, plus generous equity; a research fellowship is listed at $125K/year prorated.
Pricing
Not publicly disclosed; appears to be in beta.
Differentiators
Technology
Customers
Competitors
Estimated monthly visits
854
Theorem blog · Feb 2026
Describes Theorem’s verified translation benchmark, rocq-dove specification generation, and a 350x speedup in verified software translation.
VentureBeat · Jan 2026
Reports Theorem raised a $6M seed round led by Khosla Ventures to build automated tools that verify AI-generated software, and describes early customers and planned expansion.
Theorem blog · Oct 2025
Introduces fractional proof decomposition, which combines partial evaluation and property-based testing to scale bug finding with bug rarity.
Y Combinator
YC launch post describing the company’s trustworthy-by-default programming vision and use cases for correctness-critical software.