David Novák (Kogi): AI without hundreds of millions – quick results without a pompous transformation

In the next episode of the internal series Kogi on the edge of the knife, we talk to David Novák about how to get companies moving in the age of artificial intelligence. David Novák explains why AI is not just another technology project, but a change that in a few years will rewrite the content of most roles and divide companies into AI-native and those that will stick to traditional tools.

The biggest brake often lies not in technology or the budget for infrastructure, but in organizational paralysis and people's fear of using AI at all. And when adoption occurs only as an uncontrolled shadow AI economy, it creates individual time savings, but not a systemic change in performance, speed of decision-making, or quality of internal communication.

Watch the interview on YouTube:

In larger organizations, David Novák describes a recurring pattern: an AI agenda is created (center of excellence, AI department, AI officer), a complex strategy emerges – and the rest of the company passively waits to see what comes out of it. It is often based on the narrative: first, let's tidy up the data, then AI use cases, which in practice leads to paralysis lasting years and provides people with an alibi for not changing teams, habits, or ways of working.

Data and governance are not unnecessary. Without them, many hard use cases in risk, compliance, or automation cannot be done safely. The problem arises when the data-first approach is used as a universal excuse that halts even fast pilots where they can start immediately and with minimal investments.

The greatest value of AI may not only lie in the robotic process automation but in the soft layer of the organization: communication, interpretation, sharing strategies, meetings, minutes, emails, Teams/SharePoint, and culture. It is here that a huge loss of value arises through noise, misunderstandings, and differing interpretations, and AI can quickly help by analyzing corporate language and revealing drift from the strategy. It is also true that if the strategy is not real (lacking trade-offs, just a wish list), AI will only accelerate the production of "nice texts" that decide nothing.

Therefore, the practical first step towards managed adoption is not necessarily a massive data platform, but basic order for where the company talks and stores knowledge: where minutes, strategic documents, outputs from meetings are located, how Teams channels and SharePoint are structured. Without this, adoption will slip into a "shadow AI economy" (people use AI secretly, unsystematically), which saves time for individuals but does not move the company as a whole. And as a quick pilot, sales coaching from meeting transcripts makes sense: AI prepares minutes and specific feedback with citations and recommendations, possibly metrics – these are often necessary for management to justify the change, even though they carry the risk of "playing for score".

What else will you find answers to in the podcast?

• Why does David Novák argue that current managers will not experience a larger technological revolution – and what that will do to roles in companies

• How and why AI paralysis arises in large organizations (center of excellence, “data first,” waiting for years)

• What is the “shadow AI economy” and why does it often not bring systemic value to the company, even though it saves time for individuals

• Why it pays off to start in the “soft layer” of the organization: language, meetings, documents, interpretations of strategies

• What “minimal order” in Teams/SharePoint you need so that AI doesn’t just happen in a punk fashion

• What a practical AI use case for sales coaching from meeting transcripts looks like and why continuous feedback works

• Why sometimes without harder metrics you can't justify the change – and when on the contrary there is a risk that the metrics will destroy the essence of development

• Why even after a potential “bursting” of the AI bubble, technologies will remain and the pressure for change in work will not ease

If you want to approach AI in the organization as managed change (not as an endless strategic project or as uncontrolled “shadow”), start with a specific goal and one pilot that can quickly demonstrate value – and only then scale.