Can AI revolutionize the retail industry or is it just a modern trend?
What makes the Innovantage podcast so special? That’s the fact that for each episode its host Max Golikov invites outstanding professionals with unique expertise in business and technologies. They speak about AI and various AI-driven tools from different perspectives. This helps the podcast’s audience shape a comprehensive vision of what is happening in the tech space today.
For the 8th episode, Max, who is also Sigli’s CBDO, welcomed in his studio Žygimantas Kocius, the Chief Information Officer at Topo Centras, one of the largest retailers in the Baltics. Žygimantas has built a brilliant career. Once he started at tech positions and step by step climbed to the managerial roles. This helped him gain a deep understanding of how technologies can be used to address business needs and what businesses should do to leverage the highest value of digital transformation.
Check out the full Innovantage episode with Dr. Žygimantas Kocius here: https://www.youtube.com/watch?v=iCdYYK5hq8c&t=2342s
AI at Topo Centras
Žygimantas said that they started implementing AI in 2022, just after the launch of the first public version of ChatGPT. Everything began with the suggestion of one specialist who formulated the advantages of AI tools for the company’s business processes. The top management approved the idea. But it was impossible and even senseless to start with the introduction of a huge AI system just at once. The familiarization with AI at Topo Centras began with a couple of MVPs and experiments. The realization of such projects turned out to be not as simple as it seemed. However, today AI powers some of the company’s internal tools and chatbots.
For example, Topo Centras relies on AI-powered tools for price analysis. But at the same time, the company also combines them with solutions that do not use AI at all. This situation brightly demonstrates that sometimes it isn’t worth implementing AI everywhere just because it is a modern trend. It’s always necessary to deeply analyze the purpose and value of each tool, regardless of whether it is AI or not.
Is ChatGPT a good tool?
Probably everyone who has at least basic knowledge of what is happening in the tech world has heard about OpenAI’s ChatGPT. Despite the growing competition from other tech giants, this large language model remains the most well-known and widely used, especially among newcomers and small businesses. Žygimantas explained this popularity the following way,
Today, there are a lot of AI tools available in the market. However, it can be challenging for a small organization to test them all. So, many of them decide to use just one tool, as they expected it would likely produce similar results to the others.
Zygimantas highlighted that ChatGPT has some serious limitations. Nevertheless, it is still very suitable for some moderate or general tasks like drafting emails.
According to Žygimantas, for some specific tasks, you definitely must train your own models or at least you need to buy specific models available in the market. It’s vital to train such models with your business data, otherwise, your company will be lagging behind.
The quality of data is a key
Even if you have a lot of data for training your models but the quality of this data leaves much to be desired, you won’t achieve the desired result. As Žygimantas noted, if you are not prepared well for AI implementation and haven’t cleaned your data, most probably you will fail with any of your AI projects.
Here he mentioned one of the AI use cases that are very popular among retailers. That’s price analysis and prediction. For such tools, businesses need to purchase data and, as a rule, the quality of such data is good. But also, companies can enrich pricing tools with their internal data. And that’s where the devil hides.
If organizations tend to keep that data not in a structured way, there was no maintenance of such data, most probably the results of AI implementation will be not as expected.
AI itself can help to clean the data. For example, retail companies always deal with product descriptions and sometimes there is a complete mess in such descriptions. So with AI tools, it is possible to avoid such a situation.
Why developers should use AI tools
With his rich tech expertise, Žygimantas is sure that developers must use AI tools in their work. The reason for this is very simple. Thanks to artificial intelligence, they can save at least 20% of their time for performing various tasks. That’s quite a lot. Just imagine your yearly savings if the development costs are reduced by 20%.
Having AI at their disposal, software engineers will be able to do many more things and have more time to concentrate on more challenging processes.
But talking about AI and developers, it’s very important to highlight that AI will not work without developers.
Developers still have to read the written code and make adjustments when needed. Given this, Žygimantas doesn’t believe that human developers will disappear in the near future despite all the fears that may exist today.
Key concerns about AI
AI is an emerging technology and a lot of things about it are still not well researched. That’s why you cannot start to use it on highly sensitive data at the moment. It’s better to start implementing it in small batches with not very sensitive data. Žygimantas recommends waiting till ChatGPT and other AI tools become more well-established and mature before using them at a bigger scale in your business.
In this context, Max and Žygimantas discussed the pros and cons of using Microsoft’s Copilot. On one hand, it can be a good choice for businesses that have already joined the Microsoft ecosystem. It means that the use of Copilot won’t add any extra privacy risks. On the other hand, the feasibility of buying Copilot licenses for the whole organization can be not very feasible. The ROI is questionable, at least right now.
Another concern is related to the fact that today there is a huge number of companies that were established because of AI. They just wanted to join this growing market and hoped to get profits boosted by the hype around it. Nevertheless, 90% of those companies will die because they will not catch up. Generic AI technologies like ChatGPT will win the competition and will conquer those startups.
That’s why if you are running a business and want to buy some AI solution, you need to make sure that this solution is not provided by a two-day-old startup that just relied on the ChaGPT API.
In their discussion, Max also mentioned that the application of AI can lead to unethical decisions, for example, regarding pricing for different regions. Though it is really possible, Žygimantas doesn’t believe that it can happen at companies that have chosen an ethical path for doing their business.
According to him, it doesn’t matter what kind of technologies will be applied in the future, there still will be some companies that will act unethically. And we shouldn’t blame AI for that.
Nevertheless, there are companies that brightly demonstrate that you can do business in an ethical manner. Your transparency and your approach to customers can become your competitive advantages. When you listen to your customers, you can offer something highly valuable to them. This will be more beneficial for you than just trying to find some tricks to gain unfair profit.
EU regulation: Does it really make the market better?
The European Union is known for its strict regulations and laws related to the use of personal data. And of course, it can be viewed as a good step. Nevertheless, at the same time, it is important to understand that with such regulations countries can stop the progress.
Due to the existing restrictions, in the EU, users do not have access to the full version of ChatGPT, for example.
“But how can we regulate something that we don’t understand?” Žygimantas wondered. He said that we should monitor AI but not limit it.
To make the regulations more efficient and at the same time more favorable for the tech progress, it is necessary to track what kind of data is being used. To achieve this, regulators need to work closely with the developers of AI tools, including OpenAI, Microsoft, and others. This collaboration will be a win-win strategy for all the parties.
For some years, Lithuania was known as a fintech heaven. Thanks to the created conditions, a lot of companies were established there. They could successfully grow and develop. Now many of them are international businesses.
Žygimantas offers to do the same with AI initiatives. He explained that if Lithuania decides to become the only European nation that won’t regulate AI but will test and monitor it instead, it will be good not only for businesses but for the country itself. The money invested in this industry is huge. This could be a great benefit for the local economy.
How to structure an AI MVP
Speaking about AI adoption, Max and Žygimantas moved from the national level to companies. Max asked his guest to share his experience of implementing AI MVPs.
According to the Chief Information Officer of Topo Centras, project management plays a crucial role in the realization of such initiatives.
But there is something more that should be taken as a core thing. That’s a clear business idea. Before starting to introduce any solutions, companies should know what exactly they want to achieve.
Sometimes businesses just want to become AI-driven companies without a good understanding of why it should be done and what steps should be performed for it. In such cases, it is highly probable that they may fail.
How to implement large-scale digital transformation projects
Historically, companies were trying to implement very custom solutions in order to be competitive in the market. However, at the moment the world is going in a little bit different direction.
Žygimantas revealed his current stance. It is to be “as standard as possible”. He said that for businesses it is highly challenging to stay competitive with their own tools when there are giants like Oracle, Microsoft, or Salesforce. Such companies deeply understand the market and they are producing really good tools.
Actually, there is no reason to implement custom solutions when some off-the-shelf products with relevant functionality are already in place.
Digital transformation starts not from the systems. It should start with reviewing your own business processes and finding ways to make them as linear as possible, If you are not keen to change your existing processes, you should not even talk about digital transformation.
Then you need to decide which of the available systems is most suitable. Today there aren’t such solutions where you will not be able to make at least a couple of changes. It means that you can adjust any of the provided products to your specific needs. That’s why if a retailer is using some standard solutions, like SAP or Microsoft tools, most probably they will be extremely customized.
However, it would be wrong to say that custom development does not make sense today. The are some cases when it is impossible to avoid the necessity to write your custom code.
- For example, if you have a monopoly in your domain, the market won’t be able to offer you any standard solutions that will be appropriate for your business.
- Another reason for introducing custom products is legislation. It differs across the countries. That’s why businesses may need to implement some specific solutions that will be regulatory compliant in their regions.
- Moreover, if you are running a highly innovative AI startup, it’s highly likely that you will also be using some kind of custom solutions.
Despite the fact that Žygimantas is a proponent of standardization, he admitted that Topo Centras relies on its own custom ERP system. Nevertheless, the company has plans to replace it with a standard solution.
He explained that when a business has a custom ERP system, it can be very difficult to move away from it. Sometimes the best way to do that would be to build a separate new environment and then gradually move the features from the old system to a new one. Nevertheless, it’s vital to remember that such migration can be a rather complicated and time-consuming process. Businesses should be well-prepared for it. Every step should be planned in detail. And even some dress rehearsals conducted in advance can be a good decision.
Should we prepare for the AI future?
Žygimantas admitted that he likes science-function movies but he doesn’t believe that something of this kind can happen in reality. He strongly believes that AI will stay as a human assistant and not vice versa. He said that it is still up to us to decide how AI will be used and in what processes we are ready to implement it.
Moreover, according to Žygimantas, there will be a lot of emerging technologies coming in the near future. So there is no sense for businesses to put all of their efforts exclusively into AI.
Wrapping up
AI is a technology that can change a lot of things and processes. But we should treat it just like any other tool.
It’s important to know its limitations, understand its purpose, and bear in mind that it cannot solve every single problem. The core of everything is still the people who make the decisions.
And if you want to learn more about how we, as humanity, can benefit from the most advanced technologies like AI, we invite you to listen to other episodes of the Innovantage podcast (or to read their short summaries on our blog).
Subscribe to the Innovantage YouTube Channel
Check out the Innovantage Spotify
Listen to Innovantage on Apple Podcasts
