AI is high on the agenda of almost every organisation. Productivity, better decision-making and new forms of value creation are widely seen as the promise of the coming years. Yet in practice, AI rarely becomes structurally embedded in core processes. Many initiatives remain stuck in pilot and proof-of-concept phases. Not because the technology falls short, but because organisations attempt to accelerate without first putting their digital foundation in order.

This picture is confirmed by a recent survey among professionals who registered for our Blender Kickstart on 27 January. This event focuses on organisations that want to explore how AI can help them create strategic impact for their business.

Ambition without a foundation is not a strategy

The ambitions are clear. Eighty-five per cent of respondents cite productivity and efficiency as their primary AI objective for 2026. Improved decision-making (47 per cent) and cost reduction (38 per cent) also rank high. AI is therefore widely viewed as a strategic instrument to support business objectives.

At the same time, a large proportion of organisations are still in an experimental phase. Thirty-nine per cent mainly work with pilots and proof-of-concepts, while 30 per cent indicate they have barely started. Only 9 per cent have structurally integrated AI into multiple core processes. This contrast highlights where the tension lies: ambition is not the problem. The absence of architecture and governance is. As a result, AI remains fragmented by definition.

A telling outcome of the survey is that 67 per cent of respondents (strongly) agree with the statement that AI initiatives fail more often because a solid foundation is missing. This includes fragmented systems, poor integrations and disjointed, low-quality data, but also a lack of clarity around priorities, governance, skills and ownership. Seventy-six per cent of respondents believe that without such a solid foundation, scaling AI within their organisation is unrealistic.

Lack of skills and data quality are the main obstacles

When looking at the biggest barriers, a clear pattern emerges. A lack of knowledge and skills is mentioned by 50 per cent as the main obstacle. AI requires not only technical expertise, but also new capabilities in data management, architecture, governance and risk management.

In addition, 44 per cent indicate that their data is not in order. Without reliable, accessible and well-managed data, AI is inherently limited. It is not scalable, not explainable and not reliable enough for use in core processes.

The absence of a platform strategy (32 per cent), security and compliance challenges (29 per cent), insufficiently clear business priorities (29 per cent) and legacy integration issues (27 per cent) are also cited as obstacles. These findings show that most organisations do not lack ambition or vision, but are facing a structural maturity challenge.

Scaling without risks requires central governance

As organisations start to consider scaling AI, concerns increase. Forty-four per cent worry about security and data breaches, 38 per cent about hallucinations and the quality of AI output. Another 38 per cent see integration with existing processes as a major challenge, while 32 per cent fear insufficient governance and control.

Notably, 32 per cent indicate that they do not yet clearly understand where the biggest risks lie. This points to a fundamental issue: a significant number of organisations are scaling AI without a clear understanding of the potential negative impact on their business processes. This underlines the importance of organisations gaining a deeper understanding of AI-related risks before deploying AI indiscriminately.

The way AI initiatives are approached within organisations helps explain this uncertainty. Only 15 per cent have a dedicated AI team focused on strategy, governance and enablement. In 30 per cent of organisations, AI initiatives emerge mainly bottom-up, with individual departments or teams acting without central coordination. In most organisations (46 per cent), responsibility for AI initiatives sits centrally with IT or Digital. As long as AI remains fragmented in this way, scaling remains risky. Not because AI cannot handle it, but because the organisation is not set up for it.

AI requires a platform strategy

All of these challenges, from data and governance to skills, ultimately converge into one fundamental question: what foundation are we building to support AI in a structural way? It is essential to answer this question before even considering which AI tools an organisation should use. Organisations that continue to treat AI as an innovation or tooling issue will inevitably run into structural limitations.

At Ciphix, we see that organisations aiming to deploy AI sustainably must make the same transition: from isolated use cases to a robust digital foundation on which AI can operate securely and at scale. Only then does real acceleration become possible. The survey shows that organisations are well aware of where the friction lies. The challenge is not conviction, but maturity: the willingness to invest in choices that may be less visible, but are decisive for long-term AI impact.

During our Blender Kickstart on 27 January, this transition takes centre stage. Not the question of whether AI works, but how organisations build the foundation required to deploy AI responsibly, at scale and in a future-proof way. Register and discover how to move from isolated pilots to strategic AI impact.