Every subjective endpoint in neurology and psychiatry carries a 30–50% placebo response that drowns the real drug effect. Pharma spends $32B a year on CNS trials, and most of the failures trace to the readout, not the chemistry. That makes it a software problem, and software is what we build.
The endpoints can't tell a working drug from a placebo.
% syllables stuttered, HAM-D, CAPS-5: these scores move whether or not the drug is working. A 30–50% placebo response swamps the signal, so good molecules get killed at the readout. Change the instrument and you change which drugs survive.
This started as research, not a product hunt. Three lines of work kept landing on the same idea: you can compile what you do and don't know into rules a machine enforces. Three parts, one pipeline.
Turns scientific knowledge into rules the system has to follow. Every number traces back to a source, and the build breaks if a claim isn't backed. FORBID and ENSURE rules stop the model claiming more than the data supports.
The engine that runs those rules as text is generated. A FORBID rule isn't a prompt the model can talk its way around; it's a hard limit on what it's allowed to say. In clinical use, that's the line between a helpful guess and an answer you can stand behind.
Already running and validated on two diseases, stuttering and amyloidosis. A six-channel composite endpoint cut the placebo response from 35% to 15.4% and raised the measurable effect size by 1.3–1.4×, enough to move a trial from inconclusive to a clear read.
Experts write what's known, and what isn't, as µLang contracts with sources attached.
µLang turns those contracts into hard constraints with confidence bounds.
NIRVANA produces results that can't break the constraints.
The Clinical Framework checks every output against real data and tries to break it.
The FDA's recent guidance on real-world evidence and AI in drug development asks for the things µLang already does: every computational claim traceable to a source, stated with its confidence, with safety limits that actually hold.
We've been building to that bar. The rules are catching up to us.
Physician-trained operator who runs every business on a single principle: diagnose before you prescribe. He doesn't advise, he operates, with personal capital on the line, equity at stake, and full accountability for the outcome. Under AGG he took a pre-owned auto dealership from $250K to $1.4M per month (797th to #1 in Canada on Dealertrack in under 90 days), scaled TrustSimpli from zero to $1M in monthly revenue in 120 days, grew a BC franchise brand from 10 to 20+ locations in a year, and built Upsist, a global VA operation serving 300+ clients across 7 countries, all from personal capital with no outside investors or loans.
The scientific and technical core of Verasolve, and author of the µLang compiler (v5.1), the NIRVANA inference architecture, and the stuttering and amyloidosis clinical research that proves the platform end to end. A principal systems architect and CISO for compliance-critical systems: security-by-design and zero-trust for regulated environments (HIPAA, SOC 2, audit readiness), Rust and Cloudflare-native, and creator of HealthFees, a healthcare price-transparency engine.
Two decades building and scaling ventures across FinTech, HealthTech and MarTech. Leads a multi-layered product and growth organization at Walmart and built an agentic AI capability that sharply reduced manual marketing work. Owned a nine-figure performance-marketing P&L across Williams-Sonoma's brands and scaled QuickBooks organic growth into 12+ countries at Intuit. He's how hard technology reaches the market.
An AI-native operator who designs and runs the systems companies depend on, with one method behind all of it: diagnose the bottleneck, design the system, implement it. He doesn't hand over a strategy and walk away; he wires the AI into the operation and runs it, and consults worldwide on AI implementation and execution. With a supporting background in cybersecurity for the security-critical side of a regulated platform, his work has generated over $8M for clients across multiple industries, grown AI for Business from $0 to $3M ARR in seven months, and taken PrimeStar Solutions in Colorado from $0 to $4M in nine months: the same play, company after company.
Revenue leader whose edge is sales psychology and sales engineering: turning complex, technical products into clear buying decisions. He's sold across several industries, including medical and health, so he already knows how care organizations evaluate and buy.
Entrepreneur and executive with deep experience across construction, energy, finance and operations, having founded, scaled and led multiple companies in CFO and COO roles. His background spans financial management, accounting and banking relationships, payroll, project execution and operational teams. He specializes in building scalable systems, improving operational efficiency, and forging strategic partnerships that drive sustainable, long-term value.