In early September, Radek clarified the kinds of problems company leaders can bring to him. The result was a service catalog, published September 3, and detailed pages for each way of working together. The brief called for specific data and AI services. As the website took shape, broad statements about capabilities gave way to situations companies actually face.
What interests me as the author is how the work was divided. An independent assessment of the data serves a different purpose from a workshop, and both differ from a small AI test in one process. A company can start with what it’s missing. The three lessons below come from that work on the offering; the examples show how to apply the same thinking to your own project brief.
First, figure out whether you need reliable data, agreement, or a test
“We need better reporting” can mean several different things. Imagine a leadership team getting two different numbers for the same metric. There may be a data error. Or sales and finance may be using the same name for different measures. A new dashboard won’t decide which definition supports the decision the business needs to make.
You can see that distinction in the data audit and workshops. The audit examines the current setup and gives leadership a basis for its next decision. A workshop helps people agree on metrics, responsibilities, and next steps. The AI workflow sprint addresses a different question: is a particular process worth changing with AI?
That distinction also defines Radek’s role. He connects the technical side of data and AI to the decision a company needs to make and the people accountable for the outcome. The offering doesn’t assume every problem needs a new platform or an agent.
Try this: before choosing a provider, describe what’s keeping you from making a decision. Are you unsure whether the underlying information is correct? Do people disagree about what a number means? Or do you have a specific process and need to test a different way of doing it? These questions can overlap, but each calls for a different first step.
A small test needs a precise brief
The service catalog describes the AI workflow sprint as one manual process and a small test that informs the next investment decision. That boundary matters. A company doesn’t have to commission an entire new system just to find out whether it has the right inputs and a useful application for it.
Keeping the scope small still takes care. Consider preparing background material for a sales meeting. A demo in which AI produces an impressive summary doesn’t establish whether it helps the salesperson. Did it use current information? Did it preserve important commitments? Did it leave the person with more checking to do than the work it removed?
For a test like that, I’d recommend a mix of routine and difficult examples, someone qualified to assess the output, and agreed conditions for moving forward. If the draft needs extensive correction, that should inform the decision. Fast generation shouldn’t obscure the work needed afterward.
Radek’s format makes room for that assessment: examine the process, prepare a small test, and determine whether further investment makes sense. Finding out early that you need better inputs, a narrower task, or a simpler approach can be valuable in itself.
Write down four answers before the test: What question are we testing? Which examples will we use? Who will judge whether the result is usable? What would make us continue, change direction, or stop? Otherwise, everyone can walk away with a different meaning of “it works.”
Independent advice must leave room for no implementation
Radek’s data audit page explicitly says that further implementation work is not a condition of the engagement. An audit may recommend changes, or it may confirm that the current setup is sufficient. It accounts for the existing team or provider; replacing them is not the default conclusion.
That matters for a company considering its next investment in data. It needs to understand what works today, where the problem starts, and what to address first. Assuming from the outset that it needs new technology can mean overlooking a less expensive change to a definition, process, or responsibility.
Here, Radek’s offering is specific: an independent professional assessment of the existing data environment and priorities, with a separate decision about whether anyone should implement the recommendations and who that should be. Leadership gets a clear role for an outside partner alongside the people it already has.
Ask any provider offering this kind of work: Could the recommendation be to buy nothing new? Will our current team be able to use the findings? Will we know what needs attention now and what can wait? The answers help distinguish an independent assessment from the opening stage of selling a larger project.
When to bring Radek in
Last week’s work produced specific ways to get started for companies that need to connect a business question with data and technical options. If leadership can’t trust the numbers, it can start by reviewing the current setup. If there’s no shared brief, it can get people aligned first. If there’s already a suitable process, it can test AI’s contribution within a limited scope.
This role rests on a specific problem and a clear distinction between what we know, what we need to learn, and which next step the evidence supports. It doesn’t depend on promising that Radek can solve everything. If you’re facing that kind of decision, tell Radek about your situation. A useful opening brief can be short: what you need to decide, what’s getting in the way, and what you’ve already tried.