AI · leadership · teamwork

AI can help with work. People decide whether it helps the business.

AI can summarize information, suggest options or help draft content. Business leaders decide which problem to solve and how to tell whether the work has improved.

Start with a real business problem

Businesses differ in their work, knowledge and experience with AI. That does not mean a smaller or less experienced business should be left out. Start with three questions: What work is piling up or taking too long? Who needs the situation to improve? Who can change how the work is done?

Then ask whether AI can help with that task. Sometimes the work needs to change, not just gain a new tool. Start with a small trial whose result can be checked in real work.

AI can be wrong even when it sounds sure

AI models learn from examples and material made by people. This includes useful ways of working, as well as mistakes and bias. An answer can sound convincing but still be wrong or unsuitable for your business.

For example, AI may draft a customer email. Before sending it, a person checks whether the details are right and the tone fits. A draft is not a promise to a customer. The model does not automatically learn what your business needs from every conversation.

The amount of checking should fit the consequences. A draft may need a quick human review. A decision about money, safety or a customer commitment needs more care.

Divided work needs more team feedback

AI may finish one task faster. It does not ensure that separate pieces made by people and tools fit together. Keep the work visible and ask for feedback before a problem grows.

Agile work means taking short steps: agree on a goal, try a change, review the result and decide what to do next. Scrum emphasizes transparency, inspection and adaptation. AI can speed up parts of this loop, but the team still needs to assess the result and take responsibility for it.

When several AI agents work at once, each can handle its own task. A person still needs to bring the results together, check that they form a coherent whole and decide what to use.

Good AI use starts with clear work and good information

AI needs clear work practices and reliable information. People need to know where the data comes from, how it may be used and who checks AI responses.

In our work, a simple rhythm helps: one ordered list of priorities, a clear decision-maker, visible work, agreed checks, a cost limit and using lessons in the next step. The technical solution may be complex, but people should understand how decisions are made.

NIST’s guidance looks at AI risk throughout a system’s life. A business can use it to think through possible errors and effects, then choose checks that fit the specific use.

See what we have changed in our own work: the Ettevõtluskeskus case ↗

Let’s talk about your business

In our first conversation, we will discuss what is getting in the way, what you want to change and who can make decisions about the work. We are ready to change our first assumptions as we learn more about your situation.

Every business is different. Sometimes the first step is to clarify the work and information. In other cases, it helps to try a different way of working with one team or arrange practical training. We agree on the work and its scope after an initial conversation.

Alek Kozlov has worked on data warehouse projects with Kaubamaja Grupp, Rahandusministeerium, Baltika Group and Ettevõtluskeskus. This experience helps us pay attention to what data means, where it comes from, how reliable it is and who is responsible for it.