🎧 Listen to the latest episode of Rewire to Rise →
Behavioral Assurance
For Dynamic
AI Systems
Together with my Danish co-founders, we built Machaon to address a growing problem in production AI: systems can continue to operate while their behavior changes beyond what was originally approved.
Machaon continuously compares observable behavior against approved behavioral baselines, making meaningful change visible before it becomes operational, security or compliance risk.
Not another static checklist.
But continuous behavioral assurance for production AI.
BSL Co-Founders
Nicolai Hyllested
Architect & Technical Co-Founder: Mapping the AI Nervous System
"Detecting drift in an AI system isn't just about reading code. It's about sensing behavioral intent. We monitor the subtle micro-deviations before they become massive risks."
Nicolai is the technical architect of BSL's core innovation, Residual Stream Cartography. He brings a uniquely intuitive perspective to machine learning by blending system architecture. Mapping the internal "nervous system" of complex AI models, Nicolai translates deep-tech behavioral indicators into clear, actionable safety parameters for high-risk enterprises.
Kåre Bjørn Jensen
Strategist & Co-Founder: Corporate Governance & Scaling
"In highly regulated environments, safety isn't an afterthought. It's the foundational infrastructure. We build the commercial and legal frameworks that make trust scalable."
Kåre is BSL’s strategic engine for corporate governance, enterprise commercialization and structural scale. With deep roots in executive strategy and cross-border business development, Kåre ensures BSL’s operations are seamlessly aligned with European regulatory demands (like the EU AI Act and DORA). He opens critical institutional doors, transforming cutting-edge security research into trusted enterprise-grade assets.
Continuous Behavioral Assurance for Production AI
Machaon compares observable AI behavior against approved behavioral baselines to detect meaningful change, classify its significance and provide the evidence enterprise teams need to maintain control as AI evolves.
We Catch AI Drift Before It Becomes A Liability
Sensing internal model tension and misalignment before it translates into operational or reputational risk.
What it does:
What it is not:
Look Inside
The Black Box
See how Machaon makes behavioral change in production AI visible by comparing observable behavior against approved behavioral baselines.
Step through the interactive model to see how continuous behavioral assurance helps identify meaningful change before it becomes operational, security or compliance risk.
🎙️ Turn on the audio using the speaker icon in the top-right corner.
The Layer Around The Model
Machaon provides a behavioral control layer around production AI systems — without requiring ownership of the underlying model.
It continuously measures observable behavior against approved behavioral baselines, making meaningful change visible as models, prompts, tools, data and operating conditions evolve.
This gives enterprise teams the evidence they need to assess, investigate, approve, rebaseline or escalate changes — while maintaining behavioral assurance in production.
News & Articles
Thought leadership, analysis and perspectives exploring the forces shaping how we think, lead, work and grow —featuring expert voices, original insights and selected collaborations.
Five Paths to Smarter Digital Regulation — Without Weakening Europeans' Rights
As Brussels talks up "simplification" through the Omnibus Package, this article sets out five concrete proposals — from merging overlapping compliance rules to a European Digital Bill of Rights — for how the EU can cut red tape without cutting protection.
Magnifica Humanitas: A Tale of Four Cities
Pope Leo XIV's first encyclical reaches for the biblical tale of Babylon and Jerusalem to frame two rival visions for AI. This piece reads Magnifica Humanitas alongside the EU's AI Act — one enforceable in court, the other aimed at the moral imagination of 1.4 billion Catholics.
Digital Sovereignty — How European Is European AI, Really?
A European interface can sit on top of an entirely American stack. This piece follows the AI supply chain — chips, cloud, foundation models — to ask what "European AI" actually controls and what it would take for Europe to own enough to choose, switch and negotiate.
What AI Taught Me
About Being Human
This article explores what happens to human judgement, orientation and responsibility when we begin thinking alongside AI. Drawing on psychotherapy, lived experience and years of working with artificial intelligence, Thorsten Westphal asks a simple but increasingly important question: not only what can AI do for us, but what must we continue to practise ourselves?
FAIR HYGGE
A journey beyond the familiar Danish idea of hygge into a deeper question: what makes a human space genuinely habitable? Moving from everyday comfort through trauma, dignity and Buchenwald, the book explores whether safety and belonging can exist without requiring someone else to become smaller in order to belong.
Music as Human-Bearing Technology
An exploration of music as orientation, regulation, identity and connection - and of how AI can support human creativity without replacing the person behind it. The book asks a simple design question: what should music help a person carry, without carrying the person away?
Inspiring thoughts in your inbox!
Join the global community!
You are in. Thank you for stepping into my world!
I don't believe in perfect, surface-level stories. Here, I will be sharing the exact cognitive tools I use, right alongside my personal life, real business struggles and raw scenarios.
I am honored to share it with you.
