The courage not to chase AI. From innovation to value: where companies go wrong in implementing AI

Artificial intelligence (AI) has become something of a litmus test for a company’s modernity. If a strategy presentation lacks a slide labeled “AI”, it may raise suspicions that the company is falling behind the times. A 2025 McKinsey study shows that 88% of companies already use AI in at least one business function. However, most remain in the experimental phase, and only about one-third have managed to implement AI at scale. (1) At the same time, analyses by MIT and other academic institutions indicate that fewer than half of those investing in these technologies report significant productivity gains. A considerable number of executives admit that the benefits are difficult to measure or fall short of initial expectations. (2) The issue is not that AI “doesn’t work”. It does, and it can be highly beneficial. AI is an extremely powerful tool, and precisely for that reason, it must be used with care. As an IT partner, we at ZenIS often encounter situations where our role is not to sell yet another solution, but to say: “You don’t need this”. This article is not about why AI should be avoided. We implement AI ourselves but only where it delivers measurable economic and strategic value. And that is exactly why it is crucial to understand where it should not be applied. Strategic restraint is sometimes more powerful than technological enthusiasm.

Technology as an accelerator, not a savior

History reveals a recurring pattern: technological revolutions initially generate euphoria, followed by inflated expectations, and only in the third phase does pragmatic integration emerge. During the Industrial Revolution, mechanization did not reduce workloads it increased production speed and managerial pressure. Early electrification simply replaced steam engines without changing underlying processes; only later, when factory architecture was redesigned, did true productivity gains materialize.

AI is currently in a similar phase. It can accelerate, optimize, and analyze but it cannot define direction on its own. Without a clear direction, acceleration becomes dangerous. This aligns with the well-known Gall’s Law: A complex system that works is invariably found to have evolved from a simple system that worked.

No one has ever built a perfectly functioning complex system from scratch. First, a simple, functional structure is created then it is developed further. Yet today, companies attempt the opposite: layering AI on top of chaotic data environments, inefficient processes, and unclear accountability structures. The result? No added value, and often, losses.

One key principle should be remembered:
If a process is flawed, AI will make it faster, but it will still be flawed.

Strategic restraint as a competitive advantage

One of the most common misconceptions in boardrooms is the assumption that any process that can be automated should be automated, as this will “free up resources.” However, a critical question often goes unasked: does this process truly constrain growth or materially impact the cost structure? A practical recommendation we give clients is simple: open your P&L and identify the three largest cost categories. If the process you intend to automate with AI does not fall within these areas, and is not expected to grow exponentially, you are most likely optimizing the periphery rather than the core. Automating secondary processes means allocating strategic capital where it will not create competitive advantage. Technology must not become an aesthetic add-on that looks impressive in presentations but fails to generate real business value.

The overlooked cause of digital transformation failure

Many digital transformation initiatives fail to achieve their initial objectives. In most cases, the primary issue is not technology, but culture and change management.

Technology is the most visible element of transformation, but organizational culture - values, behaviors, and unwritten rules - determines how people respond, how quickly they adapt, and how open they are to new ways of working. If organizations lack support, clear communication, and ongoing training, transformation efforts often face resistance, uncertainty, and passive engagement.

The introduction of AI amplifies this effect. It changes not only tools but also ways of working and decision-making principles. Without proper involvement and preparation, employees perceive AI not as support, but as a threat to their competence and their role within the organization.

In organizations where stability is valued more than growth, change will be slow regardless of the technology used. Conversely, a culture that promotes learning, collaboration, and thoughtful risk-taking significantly reduces resistance and accelerates transformation.

Technology often exposes weaknesses not every organization is ready to confront them.

The workload paradox

Research published in Harvard Business Review shows that while AI significantly reduces the time required for routine tasks, it simultaneously increases organizational pace and expectations regarding productivity. (3) If a report once took two days to prepare, it can now be completed in a few hours, and naturally, more reports are expected. Without revising KPI systems, AI becomes a tool for intensifying work rather than improving efficiency.

Leaders should ask themselves: are we using the time saved to create new value, or simply increasing volume at the expense of quality?

In practice, we observe that companies failing to revise KPIs and processes alongside AI implementation often see efficiency gains “absorbed” by increased workload. In contrast, organizations that deliberately define where saved time generates higher value such as in customer experience, innovation, or strategic tasks achieve significantly greater returns from AI.

Expensive technology, limited value: where the problem lies

There are cases where AI is implemented primarily to signal progressiveness to the media, the market, or investors. However, reputation built on technological decoration without economic substance is fragile. If innovation is not tied to clear business objectives, it becomes a cost center rather than a value driver.

Consider a company that acquires a sophisticated and expensive AI tool, only to use it for basic tasks: paraphrasing text, generating simple reports, or supporting internal communication. This is akin to owning a rocket capable of reaching Mars but using it to heat water for coffee. The technology is powerful but its application does not match its potential. This is not to say such tasks should not be optimized. The issue is proportionality. If the investment does not align with its use and strategic importance, it becomes an expensive toy.

As an IT partner, ZenIS often takes on the uncomfortable role of asking questions that slow down impulsive decision-making. True partnership requires enough trust to accept unpopular answers. Only a genuine partner can say “no” because the goal is not to sell, but to build long-term results together.

Based on our experience, AI delivers maximum impact in areas characterized by high repeatability, large data volumes, and direct influence on revenue or costs. Examples include sales process analytics where even a small improvement in conversion rates can generate millions annually or risk modeling, where data exceeds human cognitive capacity. In such cases, AI is not a trend, it is a competitive necessity.

In the digital age, innovation is not the courage to adopt every new technology, it is the courage to choose strategically. Artificial intelligence is a powerful accelerator, but it amplifies the existing direction. If the direction is clear, it enables faster growth. If it is unclear, it accelerates mistakes.

The question, therefore, is not whether to use AI. The question is: where, to what extent, and for what strategic purpose. And sometimes, the wisest decision a company’s leadership can make is not to invest in the next technology, but to ask: Does this solution truly make us stronger in the long term?

If the answer is “yes,” we at ZenIS will be there to implement it responsibly and meaningfully.
If the answer is “no,” we will help identify a smarter path or build the foundations needed to move forward effectively in the future.

Article expert: Alex Schtelbe, Co-owner of ZenIS

With more than 10 years of experience in implementing innovative technologies across both the public and private sectors, Alex has worked with leading companies in the Baltics and beyond. He helps organizations modernize business processes, implement tailored e-commerce solutions, and automate daily operations to drive growth and improve efficiency.

Article co-author: Viktorija Golubova, ZenIS Marketing Project Manager

Sources:
(1) McKinsey, The State of AI in 2025: Agents, Innovation, and Transformation, 2025
(2) MIT NANDA Project Report, The GenAI Divide: State of AI in Business, 2025
(3) Harvard Business Review, AI Doesn’t Reduce Work – It Intensifies It, 2026

The original Latvian version of this article is available on the European Digital Innovation Centre (EDIC) website: https://edic.lv/lv/jaunumi-un-pasakumi/jaunumi/drosme-neskriet-lidzi-ai