Why "Non-Convex"?
In optimization, a convex problem is the easy kind. One clean minimum. One right answer. Gradient descent walks you straight there. Textbook stuff.
Real problems are not convex.
Real problems have local minima that look like solutions but aren't. Saddle points where progress stalls. Landscapes where the obvious path leads somewhere mediocre and the good answer is hiding in a direction nobody thought to look.
That's the work we do. The problems that don't have a clean slide deck answer. The ones where you need someone who has been stuck before, who knows what it looks like when a system is in a local minimum pretending to be solved, and who knows how to get it unstuck.
We named the company after the hard problems because those are the only ones worth solving.
What we believe
Most AI consulting is a confidence game. Vendors sell "transformation" and deliver a slide deck. Teams get trained on tools they'll never use. Executives buy a vision and get a proof-of-concept that dies in staging.
We think that's broken. Here's what we do instead.
Ship first, polish later. A working system that handles 80% of cases today beats a perfect architecture that's six months out. We get something into production fast, then iterate on what matters.
Be honest about what AI can't do. LLMs are powerful. They are also unreliable, expensive, and bad at math. We will tell you when a regex is the right answer. We will tell you when your problem doesn't need AI at all. That honesty saves you money and earns us trust.
Pragmatism over purity. We don't care about architectural elegance for its own sake. We care about systems that work, that your team can maintain, and that don't fall over at 2am. If the pragmatic choice is ugly, we'll write a comment explaining why and move on.
No black boxes. When we build something for you, you understand how it works. We document decisions, explain trade-offs, and make sure your team can own it after we leave. Consultants who create dependency are building a business model, not a solution.
Small teams, high trust. We don't staff projects with junior developers and a senior architect who shows up for the kickoff. The people who scope the work are the people who do the work.
Who's behind this
I'm Aaddrick Williams, and I run Non-Convex Labs.
My background is in technical program management for high-stakes manufacturing. Secure document production for government clients. Driver's licenses, passports, the documents that have to work every time.
The AI side started as personal projects. Claude Desktop for Linux (a fancy build script that repackages the Windows Electron app for most Linux distros) has over 40,000 downloads. Voice transcription plugins, CLI agent tooling. I was shipping with LLMs while most organizations were still figuring out what to put on their AI strategy slide.
Consulting work followed. I've spent a lot of time cleaning up half-hearted systems built by vendors who oversold and underdelivered. Too much talk, not enough shipping. High-stakes manufacturing calibrates your expectations about what "production-ready" actually means.
The tools I build for clients are the same tools powering my own development workflow. This site is built and maintained using a multi-agent pipeline: automated issue implementation, quality gates, spec review, code simplification, etc. We don't demo this stuff. We run it.
Get in touch
If this sounds like how you want to work, we should talk.
Get in touch