It seems like Artificial Intelligence is everywhere these days, from search engines to your refrigerator settings. Across every industry, businesses are keen to find opportunities to transform their operations with AI solutions.

But not all AI implementations go off without a hitch—successfully integrating these powerful tools requires careful research.

In this interview, Daryna Kutova speaks with Yurii Buchchenko, Head of AI Practice at Dev.Pro, about how AI is impacting operations across industries, the differences between traditional and generative AI, and how to successfully integrate these models into your business.

How is AI transforming business operations?

Yurii: Broadly speaking, AI is revolutionizing business operations by optimizing processes and enhancing decision making. When businesses get it right, the ROI can be truly substantial.

Traditional AI, for example, offers a level of precision and efficiency that manual processes just can’t match, like quickly detecting tumors from medical imaging and other raw diagnostic data.

On the other hand, generative AI is making waves in customer engagement and content generation, improving business operations that require versatile task management and creativity.

What are the specific differences between traditional and generative AI?

Yurii: Traditional AI is taught to do one thing on a deep, specialized level, while generative AI is more of a jack-of-all-trades. Generative AI can — theoretically — do anything, including the same tasks as traditional AI, though at the cost of reliability, scalability, and efficiency.

Again, one good example of using traditional AI would be cancer diagnostics. Traditional AI can use its extremely deep understanding to help doctors determine the presence of a tumor and its malignancy with unprecedented accuracy and precision, saving up to 80% of manual work time.

Generative AI, on the other hand, can identify items within images, draw pictures, and comprehend and generate audio. It’s reshaping tasks like customer personalization, providing businesses with the tools to engage customers more effectively by automatically tailoring their experiences based on their customer data.

A significant distinction to keep in mind about generative AI models is their size, however. Large Language Models (LLMs), like ChatGPT, are much larger than traditional AI models, making them more costly and challenging to manage if hosted privately.

How can businesses successfully implement AI into their workflows?

Yurii: The key to successful AI deployment lies in the alignment of AI capabilities with your business goals. You need to make sure that you have high quality, available data, and that you have the right technical expertise to manage the AI lifecycle from development to deployment.

Overall, integrating traditional AI is more complicated and time-consuming compared to generative AI. For traditional AI, gathering and consolidating data from various platforms into one usable format is a significant challenge. However, it might be the best option for your business if you have highly specialized tasks.

Generative AI, particularly pre-trained models, simplify the process and are better suited to creative tasks, but still require expertise in prompt engineering for effective use. Generative AI is also significantly more expensive than traditional models, making them more difficult to scale.

Another key point to consider is whether or not you should self-host the AI models or subscribe to third-party services. While self-hosting offers enhanced privacy and control, it demands significant resources and expertise. Third-party services provide convenience but raise concerns about data security and dependency. Balancing these factors requires understanding of business needs, risk tolerance, and company’s long-term goals.

Can AI really give businesses competitive advantages in saturated markets?

Yurii: AI can give businesses a bigger or smaller competitive edge depending on the type of AI model they implement and whether it’s suited to the tasks they require.

Many companies have data analytics platforms as a backhouse solution, providing data dashboards for human analysis — the core competitive advantage of AI is its ability to automate the decision processes based on that data. AI analyzes that data and recommends actions much faster than a human can.

Businesses operating in the retail, finance, logistics, and marketing fields can use traditional, domain-specific AI models to get ahead with predictive analytics to forecast market fluctuations and real-time insights of in-house data.

At the same time, businesses in sectors like customer service and support, e-commerce, online banking, travel & hospitality, and education are increasingly adopting generative AI to reduce labor costs with conversational AI-like chatbots.

How does the approach to AI implementation vary by industry?

Yurii: The approach depends on the nature of the business and the tasks you aim to solve with AI. Effective strategies involve carefully selecting the AI model and then tailoring the implementation process.

For industries that need AI for simple tasks, generative AI models, such as Large Language Models (LLMs), often come pre-trained, reducing the complexity of implementation. This can be a good option for customer service and support chatbots, where some creativity is required.

For more complex tasks, traditional AI is a better bet, though it requires that businesses gather and prepare vast amounts of data for training their models, demanding specialized expertise and thorough data management. Traditional AI is used in industries like healthcare, where creativity isn’t the goal; traditional AI’s ability to analyze huge amounts of data quickly and accurately is transforming diagnostics and treatment strategies, significantly improving patient outcomes.

Both types of AI provide capabilities inaccessible (or hardly accessible) to regular software products.

Have you ever seen AI implementation fail? Why did it happen?

Yurii: I’ve heard of many cases where businesses had to abandon their plans for AI integration because they hadn’t thought it through.

This happened to companies that wanted to host their own ChatGPT-like generative AI on private infrastructure, but didn’t have enough resources to do so. Similarly, I know of companies that wanted to train their own AI models using proprietary datasets, but lacked a well-defined and realistic data management strategy to fall back on.

How can businesses avoid making those mistakes?

Yurii: There are some important questions to ask yourself before diving into AI implementation. Do I want to go for a pre-trained model or train my own? Is there going to be enough data to feed into the model? Do I have sufficient resources to host my model privately, or should I do it on public infrastructure? Is there necessary AI expertise available within the company to realize this initiative? And so on.

When it comes to AI solutions, there’s no one-size-fits-all approach. A customized AI strategy is necessary to account for the specific needs, resources, and limitations of your business. This involves conducting a thorough analysis to determine the suitability and potential challenges of implementing AI solutions.


To learn more about how AI can transform your business operations, help you scale, and improve ROI, reach out to one of our Dev.Pro experts today.