Artificial intelligence is amazing — but it won’t integrate with your company on its own, you need a strategy


The enormous wave of public interest in AI is undoubtedly caused by how the results of machine learning were turned into a product bundle in the form of ChatGPT. A layman might come to the conclusion that artificial intelligence came out of the blue in 2022. Its success, however, lies in astronomically high investments in infrastructure and countless hours of human and machine labor at OpenAI. How can you make use of the currently available models in your company?

The more technical crowds have been talking about AI since the fifties, but it was first a mathematical concept — this is where we get the famous Turing test from. Thanks to brilliant scientific work, research investments, and mass adoption of computer technology in the nineties, we are now experiencing what authors in the past used to write sci-fi novels about. One could think that AI is just a software achievement. However, advancements in miniaturization, increases in computing power, and reductions in the costs of obtaining hardware play a big role too. What would’ve cost hundreds of billions of dollars back in the day now sets you back “just” a couple of millions.

The year 2022 was unique mostly in how OpenAI took this rich history and created a reliable, readily available generalized language model. But once again, it was a long and complicated process that took years of hard work, lots of money, and sufficient amounts of data, naturally. Somebody had to teach the model to do things correctly. Creating a massively popular tool from specialized technology doesn’t happen overnight. That still holds true, even with ChatGPT’s massive success and media attention.

If you’d like your company to use a similar model to deal with complicated business issues, you need to be strategic about it and put in the effort. If you don’t do that, the solution won’t work, in the best-case scenario. But in the worst-case scenario, you endanger your data, and instead of cutting costs, you’ll be paying even more and potentially dealing with legal complications. That’s why it’s better to choose an external partner to help you with implementation.

Cybersecurity, marketing, and customer care

According to research done by MIT, 87 percent of surveyed companies see artificial intelligence as a competitive advantage in their product. Yet another research by McKinsey shows that 79 percent of respondents have run into AI at some point, with 22 percent of those using artificial intelligence for work on a daily basis. But what do companies actually use AI for? A survey done by Forbes from April 2023 shows that 56 percent of the 600 respondents use it for customer support, closely followed by cybersecurity at 51 percent. Other common use cases include digital assistants at 47 percent, inventory management at 40 percent, content creation at 35 percent, and supply chain management at 30 percent.

Artificial intelligence thus saves money for companies or directly increases their revenues. In terms of savings, this is typically achieved through resource optimization, elimination of routine tasks, or reduction of personnel expenses. When it comes to the last one, it doesn’t necessarily have to mean downsizing, it can also entail that “AI supervision” allows for less qualified workers to be employed, for example. You can theoretically factor in savings from damages and errors that will be prevented by artificial intelligence. We’ve also observed further applications, especially for tech companies. We see improvements in product user experience among our clients and partners, automation of internal processes, assistance with employee selection, personalized training, and last but not least, a better overview of company data.

As for increases in revenue, there is the potential for improved marketing or better customer support, which brings the company’s name to a wider audience. In product-driven companies, it’s also about added value, such as a new feature that makes the product easier to use. Nowadays, products have clearer and more interactive knowledge bases, text generation fields, and various automation functions. This can bring in new revenue through increased popularity among developers. This can then result in broader product adoption in other solutions, a launch of a new premium subscription, or attraction of new customers who are comparing multiple products. While this sounds great, even the best innovation won’t just happen on its own.

There are risks, but they can be managed

As I said before, implementing artificial intelligence isn’t an easy process. You can run into complications with organization, competencies, and the technology itself. On top of that, there are of course legal and ethical considerations as well.

Organizational challenges lie in the fact that innovations must have support from the company’s leadership. Otherwise, it becomes very difficult to navigate. For our clients, for example, we offer full-day workshops in collaboration with FEB Ventures and Alpha Industries. We assist C-level management in aligning themselves with the goal of implementing AI within the company and determining the appropriate approach. We perceive this as crucial — the potential of any technology won’t be successfully utilized if you don’t understand it and if you don’t set up goals.

Competency issues are related to educating employees about the capabilities and limitations of AI. Emphasis on training and the intuitiveness of solutions is key here. We have addressed this point internally with our own guidelines, providing everyone with clear boundaries regarding which tasks can be delegated to the model.

Technological obstacles may arise if AI tackles a problem on the edge of its abilities or even beyond. Such situations can lead to failure or higher costs, which can be addressed by conducting a feasibility study before investing significant resources into implementing the solution. For certain specific tasks in your company, the technology might simply not be there yet, and its deployment will only make sense in the future.

Legal issues can pertain to intellectual property or data privacy. These areas, especially due to the current rapid adoption of artificial intelligence, are widely discussed nowadays. Considering the upcoming EU directive on AI usage, it’s important not to underestimate the situation and to consult this domain with experts.

Ethical issues are rooted in the values and principles that artificial intelligence should uphold in the context of your company. Examples include transparency, privacy protection, and contributing to employee satisfaction. For instance, we’ve introduced a principle that says that AI should be used in a way that positively impacts the society around us. This serves as a compass of sorts for potential innovations.

Develop or utilize an existing solution? 

You are most likely already using artificial intelligence at work. If you are considering broader implementation within your company, I definitely recommend it in some form. It doesn’t have to be about new products or features that come with a lot of in-house development. You can start experimenting in a testing environment, perhaps with open-source tools. Essentially, it’s good to thoroughly explore the technology, its capabilities, and limitations and apply them to problems resembling your specific company challenges. The first solution you reach for might not satisfy your needs right away. The AI world is not just ChatGPT, which receives attention in both positive and negative contexts. Our real-world experience has enabled us to put together a process for considering such a step and not getting burned in the process.

  1. Identify opportunities 

Examine the routines and tasks that burden you and your teams and how they could be automated. Similarly, get some feedback from your customers and users in relation to your product or service. Then evaluate your cost structure and explore optimization opportunities. Consider how your product or service could actually progress. Do you want to add more value to current clients? Do you want to reach a broader audience? Or perhaps both?

  1. Set clear goals 

Remember to set goals in advance. These should be clearly measurable and sufficiently understandable. What do you actually consider as success? What should be the quality and reliability of the system? How much money will the system save or bring you? Answering these questions will also make it clear whether it’s worth developing some form of your own AI solution or if using existing tools will suffice.

  1. Determine priorities 

Prioritize the opportunities and stimuli you identify based on how much positive impact they will have on you, your teams, your product, or your company. However, don’t try to generate the largest number of ideas. Instead, focus on specific areas and deeper benefits.

  1. Seek advice 

Get in touch with AI experts who can provide you with technical expertise and an external perspective. The broader context and experience are invaluable in this regard. Don’t forget about specific challenges in areas beyond technology, such as law or ethics, and consult experts. Also, don’t hesitate to ask for input from your employees.

  1. Choose the right path 

Based on this process, you will know how to utilize existing AI tools, add AI directly to your product, or create an entirely new product.

Image credit: Laurent T / Shutterstock

Filip Kirschner is co-founder and COO of Applifting.


Leave a Reply

Your email address will not be published. Required fields are marked *