AI has yet to live up to its promises. After being billed as a panacea for the business world (a Harvard Business Review study shows that 89% of business leaders believe that AI will be the most transformational tech in a generation), many adopters expected an overnight transformation. When that didn’t arrive, disillusionment set in.
It’s easy for those who bought into the hype to throw their hands up in frustration and consider scrapping initiatives. However, organizations that develop a pragmatic approach to AI implementation will not only celebrate small wins in the short term but also set themselves up for bigger victories down the road.
Generative AI may seem like an afterthought with agentic AI on the mind and physical AI on the horizon, but many businesses have yet to conquer and replicate the adoption of any AI. As long as we’re being practical, it’s not too late to get started.
So, how can a business remove its rose-colored glasses to pursue a more realistic approach to AI?
1. Abandon short-term expectations, ye who enter here
Many AI initiatives begin with pie in the sky aspirations, but tempering expectations is essential to avoid being let down in the early stages of any AI journey. Those hoping that workflows would be instantaneously automated, and workloads lessened for employees, had a rude awakening.
In fact, one Upwork study shows 77% of employees using AI believe the tools have increased their workload, while 47% of those using the tools claim they do not know how to achieve the promised increase in productivity. What started as a project to reduce workload paradoxically creates new burdens: data scientists spend countless hours fine-tuning prompts, engineers struggle to verify outputs, and domain experts must meticulously review AI-generated content for accuracy.
Instead of shooting for the moon with AI solutions, it’s important to have practical ideas of where implementation can have the most impact. Ignoring the hype to find less flashy success with initial AI programs shows value to employees and stakeholders, rather than leaving them disappointed when big dreams fail to materialize.
2. Don’t start the game without a playbook
At every level of team competition, from Little League to the pros, coaches put together a playbook to increase their chances of victory. The same must be done by businesses looking to implement AI. Developing an AI playbook begins with assessing where AI will make the most impact to find a clear path forward.
Strategic implementation allows AI systems to learn, improve, and deliver effective solutions. For instance, if an independent software vendor (ISV) wanted to transform legacy code migration, it could turn to Gen AI and achieve cost-effective and high-quality results. The ISV’s platform learns and refines its processes through reinforcement, learning from human feedback. This reduces migration time and eliminates dependency on outdated systems.
As NVIDIA CEO Jensen Huang pointed out: “The excitement around AI agent systems that can reason, plan, and act more autonomously was front and center at NVIDIA GTC 2025. But so was a reality check: these models require significantly more computational power to function effectively. In other words, the future of AI may be smarter, but it’s also more resource intensive.”
This is where expectations need to be recalibrated – not just for potential outcomes, but for the time, infrastructure, and investment needed to get there. Improving back-end processes like this may not grab headlines for innovation, but it provides tangible ROI and optimizes essential processes. Such an iterative, strategic approach to AI implementation is essential across industries to score real-world wins that lay the groundwork for further adoption and eventually turning bigger dreams into reality.
3. Address the AI elephant in the room: data
Don’t expect organization-wide transformation to happen overnight. It takes time to teach employees how to work with AI platforms, as well as to prepare existing systems and workflows for the new tech. Workshops ensure that employees working with the AI platform know how to use it effectively and have a clear understanding of what the tech is meant to accomplish.
Even as advanced AI solutions become easier to implement, many businesses find that their data is not yet ready to take full advantage of the solutions’ power. Real-time data accessibility and analysis is essential for smooth information exchange with AI.
If a company in the industrial sector wanted to use AI to minimize equipment downtime, it would need the platform to interact efficiently with IoT devices and machine documentation to notice changes in patterns. But that isn’t possible with unstructured data.
A solidly defined data infrastructure allows an AI solution, like the one sought by the hypothetical industrial business, to use historical data analysis and synthetic data generation. The AI can then assess the health of machinery and calculate the ROI on repairing versus replacing equipment.
Good data architecture also makes it easier for employees to work alongside the AI. This brings home some of the benefits of AI to users as they see their work become easier. Chaos reigns in the absence of a strong data foundation.
At the same time, resentment festers among the aforementioned employees reporting increased workloads after AI solutions are implemented. One study found 58% of business leaders claim their organization uses inaccurate or inconsistent data most of the time, if not always. The same study discovered 65% believe no one at their organization understands how the data is collected or how to access it. Both findings are problematic.
But data – with or without AI – is essential for every business. Prepping data for early AI projects paves the way for success, both in the short-term with initial projects and in the long-term as AI initiatives become more complex.
4. Accept that AI success requires a switchboard, not a light switch
While some organizations have yet to reach the efficient utopia they envisioned when they first began their AI journey, there are still victorious conquests to be had. Companies must navigate through the promises and pitfalls of AI with practicality.
Collectively thinking of AI as some magical solution is misguided; a more nuanced understanding of AI as a tool to complement and not compete with existing systems is required. Adopters must focus on strategic implementation and identify areas where AI can have the most impact, instead of reaching for the stars with every initiative. The sooner businesses see AI as the vessel for augmenting human capabilities – rather than replacing them entirely – the easier these ugly truths and frustrating hurdles become a fleeting memory of AI’s past. Once realistic goals are set and data is prepped, the promise of AI is suddenly much more attainable.
