Companies racing to deploy AI are discovering that project success often comes down to the quality of their data, leaving many initiatives stuck at the proof-of-concept stage and never delivering tangible revenue.
Martin Frederik, regional leader for the Netherlands, Belgium, and Luxembourg at data cloud giant Snowflake, offered a blunt diagnosis in an interview. “There’s no AI strategy without a data strategy,” Frederik says simply. “AI apps, agents, and models are only as effective as the data they’re built on, and without unified, well-governed data infrastructure, even the most advanced models can fall short.”
That pattern plays out in many organizations: an early proof-of-concept wows internal teams but fails to become a money-making tool. Frederik argues that a lot of the time leaders mistake the technology for the end goal. “AI is not the destination – it’s the vehicle to achieving your business goals,” Frederik advises.
Projects that stall tend to share familiar problems. The work isn’t truly aligned with business priorities, departments don’t communicate, or the underlying data is inconsistent and messy. Statistics suggesting that 80% of AI projects never reach production can feel discouraging, though Frederik offers a different take: many of these setbacks are “part of the maturation process.”
When the foundation is laid correctly, returns can be significant. A recent Snowflake study found that 92% of companies are already seeing a return on their AI investments. In fact, for every £1 spent, they’re getting back £1.41 in cost savings and new revenue. Frederik keeps returning to the same prescription: start with a “secure, governed and centralised platform” for your data.
Technology alone won’t carry an AI program across the finish line if the organizational culture isn’t prepared. One major barrier is getting data into the hands of everyone who needs it, not only a small group of data scientists. Making AI work at scale requires firm foundations across “people, processes, and technology.” That calls for breaking down departmental walls and making accurate data and AI tools broadly accessible.
“With the right governance, AI becomes a shared resource rather than a siloed tool,” Frederik explains. When teams operate from a single source of truth, disputes over whose figures are correct drop away, and decision-making speeds up.
A significant shift on the horizon involves AI agents that can reason across diverse data types regardless of how neatly that data is structured. These systems can move between tidy spreadsheets and the unstructured content in documents, videos, and emails. Considering that this unstructured data makes up 80-90% of a typical company’s data, the capability marks a major step forward.
New interfaces let staff with little technical background pose complex questions in plain English and receive answers directly from the underlying data. Frederik describes this evolution as “goal-directed autonomy.” Until recently, AI tended to act as an assistant that required constant prompts. “You ask a question, you get an answer; you ask for code, you get a snippet,” he notes.
The next generation of agents will behave differently. Given a complex objective, an agent can map out required steps on its own, from writing code to pulling information from other applications, and then return a complete result. This automation will take over the most time-consuming aspects of a data scientist’s role, including “tedious data cleaning” and “repetitive model tuning.”
The practical effect is to free technical staff to concentrate on higher-value work. That shift raises employees up “from practitioner to strategist” and lets them focus on driving measurable impact for the business.
Snowflake is a key sponsor of this year’s AI & Big Data Expo Europe and will send speakers to share practical insights during the conference.