The Impact of Non-Autoregressive AI on Critical Systems
Autoregressive
Non-Autoregressive
LEAN4 Verification
Non-autoregressive AI models represent a fundamental opportunity for the future adoption of AI in critical systems. Autoregressive models generate outputs sequentially, step by step, while non-autoregressive models evaluate entire solutions at once, optimizing globally across constraints.
The result is advanced mathematical reasoning, which, when combined with a LEAN4 proving lens, allows formal verification at unprecedented speed and accuracy. This approach applies directly to crypto, automating formal verification of smart contracts and blockchain protocols that previously required months of manual auditing.
The same methodology scales to other critical areas, including high-frequency trading, semiconductor design, and aerospace systems. This deeply technical work is led by engineers with rigorous backgrounds including quantum physics, reinforcement learning, ICPC competitive coding, and mathematics.
Experiencing this firsthand as a New York-based go-to-market lead at Logical Intelligence in Fall 2025 shaped my understanding of how these architectures are defining the next frontier of AI.
The race isn't between humans and AI. It's between AI that can only predict and AI that can prove. The difference might be everything.
Harvard Business School
Research Associate under Professor Rosabeth Moss Kanter, 2023–2025. Co-authored 20+ published case studies on sustainable building, infrastructure, and climate innovation.
EcoFi
$100K → $6M revenue in 18 months
Climate Pledge Arena
$1.15B · 100% renewable · 92% waste diverted
Highland Electric Fleets
$38 / charge vs. $180–$240 diesel · LA28 sponsor
Can Cities Beat the Heat
14 U.S. cities · comparative climate action analysis
Miccosukee & the Everglades
3M+ people rely on the restored water system
Vineyard Wind
$4B · 800 MW · ~400,000 homes powered
Blue Frontier
HVAC = 3% global emissions · 85% more efficient
GE Appliances
Net Zero Neighborhoods · heat-pump water heaters · smart grid
Climate Action in Miami
Rising seas · Citadel · living seawalls · tech hub blueprint
Carbon Capture
36,000 t CO₂ /yr removed · ~$1,000 / ton · DOE DAC Hubs
Green Buildings
75% less energy · 75% less water · 85%+ renewable electricity
Theses in Progress
One Way I'd Revitalize New York
What if a sleek crowdrising app mapped every available urban works project across the five boroughs — each with a rendering of what it becomes, what it costs, and how close it is to funded? Users vote or donate. City approval triggers the funds. The crew ships.
StreetEasy changed how New Yorkers find housing. Citizen changed how they understand safety. Worldchangers could change how they build it.
One Way I'd Revitalize Buffalo
Buffalo has a low cost of living, a rising tech workforce, and proof of startup momentum in ACV Auctions and 43North. What it lacks is repeatable venture infrastructure — a founder-led studio that pairs validated ideas with local operator talent, deploys MVPs with pre-seed capital, and builds companies that stay, hire, and scale in place.
I wrote the platform thesis. I think it works. I think it works in Rochester, Pittsburgh, and Syracuse too.
"Cities are the most underrated technology platform on earth."
"The next great record label will be run by someone who thinks like a VC."
"Formal verification is going to change security the way containers changed software."
Making Things
A music app I've been designing.
22 years of guitar. Writing songs in New York.
Curriculum Vitae
Commercial interior construction management, Manhattan. Corporate tenant build-outs, medical spa fit-outs, CSI division estimating.
Under Professor Rosabeth Moss Kanter. Co-authored 20+ published case studies on sustainable building, HVAC, climate tech, and infrastructure.
Go-to-market strategy for non-autoregressive AI applied to formal verification of smart contracts and critical systems.
Minor: Energy Science & Policy. Study Abroad: Barcelona, Spain, Spring 2022.
Full CV available upon request — jacobasmall@gmail.com