I care about what
technology does
to people.|
Not only what it does for them.
What I do.
I'm an innovation consultant. Twenty years across technology, research, and strategy — most of it in healthcare and life sciences. I help companies decide what's worth building, and how it will create and capture value.
AI has been part of my own work for a few years now. I noticed it changing my work, and me. So I started researching how it changes knowledge work and what it does to the rest of us. That research is Modrn Mind, below.
What I'm figuring out.
I started using LLMs when ChatGPT 3.5 launched in late 2022. Soon after, I noticed I couldn't recall details of my AI-assisted work. I was seeking confirmation from AI on things I was good at before. I was losing my ability to think on my own. Then I realized this wasn't just happening to me.
Modrn Mind (yes, no "e") is my research into what AI does to those of us who think for a living. I look at how AI use affects our agency, judgment, and relationships, and what keeps our work valuable as AI does more. It's an attempt to help people notice this pattern in themselves and do something about it. The research lives in an open knowledge base on GitHub, and it's becoming a book.
Explore the Knowledge Base →- Charting the Course: Strategic Transformation for Software Agencies in the AI Economy why software agencies have to move from commodity delivery to the value only people can create
- The Human Edge: Cultivating the Human Wisdom AI Can't Automate as AI takes over the craft, the premium shifts to the human wisdom it can't copy
- The One Calibration Question That Changes How You Use AI before each prompt, ask how much of you the task actually requires
- Why AI Productivity Feels Like a Trap (and What to Do About It) your output gets better while you quietly get worse, and how to tell the difference
Twenty years.
Still curious.
Different fields, but it's the same problem I still work on. New technology only pays off when the people and the business move with it.
// Inventing products
Detecting autism from the way a child plays
Children play a game. The game captures fine motor patterns: timing, pressure, movement. ML models analyze those patterns for early signs of autism. A diagnostic tool disguised as play, designed for children too young to answer questions.
Turning a phone into a blood lab
Miniaturizing satellite hyperspectral imaging into a smartphone dongle. Point it at a blood sample, create a digital spectral cube, transmit it to the cloud for analysis, receive a diagnosis. Designed for places with mobile connectivity but no laboratory infrastructure.
Rapid STD testing in a ring
A wearable medical device with a replaceable cartridge containing microneedles and a biochip, testing for four common STDs in minutes. No clinic, no lab, no waiting.
// Proving it works
From hypothesis to Phase 3 clinical trial
Starting with exploratory studies. Refining the ML models. Publishing in peer-reviewed journals. Replicating results. Then a Phase 3 clinical trial in 760 children across the UK and Sweden. A new diagnostic paradigm needs to be proven before anyone will trust it. Building that proof, one study at a time.
Will the model hold in the real world?
ML tools trained in controlled research settings perform differently when deployed in real environments. Different devices, different lighting, different operator behavior. Predicting and managing that gap before it becomes a product failure.
Building a global research network
Setting up and maintaining collaborations with universities, research centers, and therapeutic centers across multiple countries. Each with their own agendas and timelines. Building the agreements to make joint research possible and align interests.
// Making new things pay
A business model for something that doesn't exist yet
Building a commercial model for a digital health product with no category in any healthcare reimbursement framework, no established buyer, and nothing comparable to point to.
How to create value before the science is complete
Scientific and clinical validation takes years, leaving the question of how to start being useful to users, building trust with practitioners, and staying financially alive while the evidence base is still being built.
When clients want ML but aren't ready for it
Companies arrive asking for machine learning. What they actually need first: clean data, infrastructure to move it, people who know what to do with it. Building the foundations that make ML possible before ML itself.
// Changing how companies work
Selling something a sales team doesn't understand
Getting a sales organization to sell data and ML services when they don't understand what they're selling, and their bonuses depend on services they already know how to close.
Turning a software company into an advisory firm
Shifting a company that competes on execution and price toward competing on value and results. Different positioning, different talent, different sales process, different client relationship.
Staying fast while becoming certifiable
ISO 13485 demands documentation, process control, and audit trails. Getting certified without turning the company into something slow and risk-averse. Building the quality system around maintaining speed and flexibility to explore new ideas.
Some of the work got noticed.
MIT Innovator Under 35
Named by MIT Technology Review for work on early autism screening using AI and games.
New Europe 100 Challengers
Selected by Google and the Financial Times as one of New Europe's 100 most innovative changemakers.
Singularity University Fellow
Nine weeks at NASA Ames Research Park, working with leaders from around the world on humanity's hardest problems.
TEDx Speaker
Spoke at TEDxKrakow on entrepreneurs using business as a force for genuine good.