We Built Bollwerk Because We Lived the Problem
We’re Arkadij and Oisin. We’ve been solving hard problems together since high school.
In 2021, we developed an investment thesis: the Covid-fueled stock mania pushed the low-quality or outright fraudulent companies to historical highs. It was the perfect time to start shorting junk. The first usable LLMs were emerging at the same time, and we thought it would be fun to try and beat activist short-sellers at their own game.
We set out to build a single, sophisticated model to detect fraud in public companies, using the same obscure data sources they do; from Japanese securities filings to European hedge fund disclosures.
The Barrier
The AI model was the easy part. The real nightmare was the infrastructure. We spent 80% of our time wrestling with the basics: sourcing and cleaning the data, versioning prompts, securely deploying the agents, and then trying to share any useful insight without it being a massive operational lift. We were drowning in bespoke tooling and duct tape.
The Core Insight & Platform
That's when we realized we'd solved the wrong problem. The prize wasn't our fraud model; it was the engine we'd been forced to build to run it.
We looked at the mountain of infrastructure we'd built and saw the real challenge every financial institution faces: It’s not a lack of models, it's the crippling difficulty of deploying and managing them in a high-stakes, secure environment.
So, we productized our engine. That's our platform. It gives our clients the one thing we desperately needed: a secure, standardized way to deploy AI-powered risk applications without having to build everything from scratch. We provide the rails so you can focus on the risk, not the plumbing.
Oisin
Arkadij
Our Operating Principles
Our experience building high-stakes financial models taught us a hard lesson: great AI is useless without production-grade infrastructure. The principles below aren't just beliefs; they are the engineering and security commitments we built into the core of our platform.
Our core values
The guiding principles
that define us
No Black Boxes
Your team needs to trust the outputs. Our platform is built for explainability, providing clear audit trails for data lineage, model versions, and prompt histories. You get control and transparency, not a magic box that could become a liability.
Production-Grade Reliability
Prototypes don't survive in the real world of finance. We moved beyond fragile, bespoke scripts to a standardized, robust framework for a reason. Our platform is architected for the uptime, scalability, and rigor required when millions are on the line.
Pragmatic, Not Theoretic, Innovation
The goal isn't to use the trendiest new model; it's to solve a business problem. We integrate cutting-edge tech where it delivers a measurable edge, but we are grounded in the practical reality of what it takes to run, manage, and get value from these tools every day.
Zero-Trust by Design
Security is our foundation. We learned the hard way how many vulnerabilities exist in a typical AI workflow. Our platform is built on a zero-trust security model and deployed within your environment, ensuring your data and models remain your own. Full stop.
Learn about the platform
Explore how our AI-driven platform can boost and streamline your operations.
Why Bollwerk
Keep up with the latest
Stay updated with the latest developments from our team by checking our blog.
Read Blog
Join the team
Become a part of our mission —explore career opportunities at Bollwerk AI.
Career