AI Ambition vs Reality: What a Cloud and Data Expert Sees in Practice

Over the past year, artificial intelligence (AI) budgets among large companies increased from $89 million to $114 million. At the same time, 74% of organisations now rank AI among their top three priorities, up from 60% a year earlier. This growing investment reflects rising pressure to act and a belief that AI will play a key role in future business success.

However, Robertas Skardzius, a Cloud and Data solutions specialist at Baltic Amadeus, notes that while ambition is growing fast, technical readiness often stays behind. In this article, he explains where companies most often overestimate their AI readiness, why data quality matters more than tools, and what practical steps help turn AI from an idea into real business value.

AI is still seen as a productivity shortcut, not a business engine

Many companies still see AI mainly as a personal productivity tool rather than a driver of wider business change. It is often linked to everyday tasks like writing emails faster, searching for information, or polishing text with tools such as ChatGPT.

‘People tend to think of AI as a smarter search engine or a writing assistant,’ Skardzius explains. ‘Not as something that can reshape processes or products.’

This view captures only a small part of AI’s potential. The real value appears when AI helps organisations automate processes, reduce manual work, and make better decisions across teams.

At the highest level of maturity, AI becomes part of the product or service itself. It can generate insights, predictions, or personalised recommendations that directly influence revenue, from dynamic product suggestions to customer behaviour forecasting.

‘This is where AI stops being about saving time and starts being about making money. AI allows companies to respond dynamically, instead of relying on fixed rules,’ he adds.

AI is only as good as the data behind it

‘AI does not create knowledge on its own. It works with the information it is given and simply delivers results much faster than a human would. If a person would make a wrong decision based on poor data, AI will do the same, only faster,’ explains Skardzius.

He compares data quality to fuel in a car. Poor-quality fuel may not stop the engine immediately, but performance suffers, and over time, the damage becomes unavoidable. The same applies to AI. When data is inaccurate or inconsistent, outcomes become unreliable, no matter how advanced the technology is.

To make it simpler, here are five core principles that help determine whether data is fit for AI use:

- Correct – the data reflects reality and contains accurate values.

- Sufficient – there is enough data to support meaningful analysis and decisions.

- Relevant – the data relates directly to the problem the AI model is expected to solve.

- Consistent over time – data is complete and continuous, without gaps that distort results.

- Understandable – each data field is clearly defined and structured so that both people and systems can interpret its meaning and purpose.
 

Skardzius observes that while teams actively follow market examples and learn from others, including insights shared by Baltic Amadeus, this curiosity often stops short of data readiness. In mindset, companies are ready to use AI. In practice, many are still catching up when it comes to preparing and understanding their data., this curiosity often stops short of data readiness. In mindset, companies are ready to use AI. In practice, many are still catching up when it comes to preparing and understanding their data.

Cloud Data platforms and Microsoft Fabric

Another common challenge in AI projects is data fragmentation.

‘Data may be correct at the source, but as it travels through different systems, its quality can degrade. By the time it reaches the AI model, the result may already be unreliable,’ explains Skardzius.

This is where cloud data platforms become especially valuable. Centralised cloud data warehouses help avoid the so-called ‘broken telephone’ effect by ensuring that data is collected, cleaned, and accessed from a single, trusted source. This not only improves reliability but also shortens the path from idea to implementation.

Public cloud is not mandatory for every AI initiative, but it often makes the first results easier and faster to achieve. Cloud providers such as AWS, Azure, and Google offer ready-to-use platform services that remove the need to build and maintain everything from scratch. This allows companies to experiment, scale up or down, and even run short-term AI initiatives without long-term infrastructure commitments.

Modern platforms such as Microsoft Fabric take this a step further by bringing data engineering, analytics, and AI capabilities into one unified environment.

‘Previously, companies had to assemble separate tools and then invest time and effort to connect them into one system,’ he says. ‘Today, unified platforms allow organisations to move much faster, from preparing data to actually testing AI use cases.’

This kind of setup also makes it easier to understand where data comes from, how it is used, and how decisions are made. For many organisations, it becomes a natural extension of their data analytics efforts and a more reliable foundation before moving deeper into AI-driven solutions.

Final thoughts: AI starts with clear goals, not with technology

According to the expert of Baltic Amadeus, successful AI initiatives always begin with a clear business goal. Before thinking about tools or platforms, companies need to understand what they want to improve, whether that is efficiency, customer experience, cost reduction, or revenue growth.

The next step is defining how success will be measured. Without clear metrics, AI initiatives risk remaining experiments that sound promising but deliver little real impact.

Only after that does technology come into play. At this stage, data collection, cleaning, and structuring become unavoidable parts of the journey. Without this groundwork, even the most advanced AI solutions struggle to deliver reliable results.

‘AI will not replace strategy. But without strategy, and without data, it will not deliver value either,’ Skardzius concludes.

In the photo: Robertas Skardzius is a Cloud and Data solutions specialist at Baltic Amadeus