Define what matters.
Set the product vision, architecture and constraints. Decide which outcomes are valuable and what evidence would make a solution acceptable.
HUMAN EXPERTISE. EXPONENTIAL POTENTIAL.
Human attention
is the bottleneck.
Put it where it matters.
AI can generate more work than we can give our attention to. The opportunity is to build a system that puts expert judgment at the right moments — and lets AI carry the work forward.
Built with your people. Around your processes. In your tooling.
THE OPERATING MODEL01 / ∞
HUMAN JUDGMENTSet the direction.
intent + constraints
AI EXECUTION LOOP
ExploreChallenge & propose
BuildGenerate & refine
VerifyTest & gather evidence
Iterate within agreed boundaries
evidence + exceptions
HUMAN JUDGMENTMake the decisions that matter.
Expertise stays in the loop.Repetition moves out.
VISION LED BY PEOPLEEXECUTION AMPLIFIED BY AIACCOUNTABILITY STAYS HUMAN
01 / THE CONVICTION
Attention is scarce.
Treat it that way.
We keep asking how much AI can do.
I think the better question is:
where do we still need a human?
Human attention is one of our scarcest resources. Spending it on every intermediate step limits what even our best people can achieve.
AI should challenge our assumptions, brainstorm with us and generate solutions. Humans should define the product vision, shape the architecture and make the trade-offs that determine whether we are building the right thing.
A dark factory puts those decisions at the centre of a repeatable delivery system. The ambition is for one expert to create exponentially more value by directing their attention to the places where it has the greatest impact.
The factory does the work.
The experts make it matter.
EXPERTISE IS THE FOUNDATION
A dark factory
isn’t vibe coding.
It cannot operate without specialists.
Generating code is only part of delivering software. Someone has to understand the domain, choose the architecture, define what good looks like and recognise when the result is wrong.
In a dark factory, that expertise shapes the system: its context, constraints, feedback loops and quality gates. Specialists decide where autonomy is useful and where human judgment remains essential.
Less attention on repetition.
More attention on consequence.
02 / FROM INTENT TO EVIDENCE
An operating model.
Not just another AI tool.
A dark factory connects the work from product intent to delivery. Each stage has a purpose, clear boundaries and a way to know whether it worked.
Challenge. Explore. Build.
Use AI as a sparring partner and an execution engine: question assumptions, compare approaches, generate solutions and iterate against feedback.
Verify before moving on.
Turn expectations into checks. Gather test results and reviewable evidence. Route consequential decisions and exceptions back to the right specialist.
Feedback improves the product, the blueprint and the next iteration of the factory.
03 / WORK WITH ME
Your team.
Your tooling.
Your factory.
I help organisations set up a dark factory, working alongside their technical teams. I bring my experience and blueprints; together, we adapt them to the way your organisation actually works.
Talk through your starting point- 01
Understand your reality
Map your delivery process, repositories, tooling and bottlenecks. Find where expert attention is being spent and where it creates the most value.
Outcome: a clear starting point and a focused use case. - 02
Adapt the blueprint
Design the workflows, context, integrations and quality gates with your specialists. Fit the factory to your processes and tooling.
Outcome: an operating model your team understands and owns. - 03
Put it to work
Build a bounded pilot around real work. Test the whole path, make the evidence visible and refine where the system needs expert intervention.
Outcome: a working factory in a defined scope. - 04
Make it your capability
Work hands-on with your people, share the blueprints and help them operate and improve the setup. Expand autonomy as the evidence supports it.
Outcome: a capability that belongs to your organisation.
EXPERIENCE, APPLIED.RADEK KOWALSKI / AI AUGMENTED TEAMS
I build across engineering,
architecture and leadership.
I bring over twenty years of experience across these three worlds. I’ve led engineering teams, shaped architectures and built AI-powered products in production.
I work with both the people setting direction and the people delivering it. A dark factory needs that connection: product intent, sound architecture and a team that knows how to make it work.
More about AI Augmented TeamsA FEW PRACTICAL QUESTIONS
Let’s be clear.
Does “dark” mean no human involvement?
No. It means the repeatable work can run without constant human attention, inside boundaries set by specialists. People own the vision, architecture and accountability. The amount of autonomy depends on the work and the evidence.
Is this about replacing our specialists?
Specialists are essential to making it work. The purpose is to multiply what they can achieve by moving their attention towards judgment, product direction and the hardest problems.
Do we have to replace our tools and processes?
The starting point is your existing environment. We work out which parts to connect, which practices need to evolve and where a small change will have the greatest effect. The factory should fit your organisation.
Where would we start?
With a conversation about your team, your delivery process and one concrete bottleneck. From there, we can identify a bounded use case, define success and decide whether a pilot makes sense.
PUT YOUR EXPERTISE TO WORK
What could your team do
with its attention back?
Tell me where delivery gets stuck. Let’s explore what a dark factory could look like in your organisation.
aiaugmentedteams@gmail.com