Why are companies struggling to implement AI in meaningful ways?

AI has found its way into almost every industry, from agriculture to transportation and everything in between. But while there’s no shortage of companies who have begun to explore the technology’s potential, there is a shortage of companies that have successfully implemented it into their business processes in any meaningful way. 

In this episode of Qcast, we’ll understand are challenges to implementing AI in meaningful ways and what can companies do to address them.

Sahil Brij Malhotra: Hi I am Sahil Brij Malhotra, director of business development at Qualetics Data Machines. Welcome to Qcast, where we bring in our leaders, experts, and guests to talk about AI challenges and opportunities. Today we are going to interact with Sumanth Vakada, founder and CEO of Qualetics and we are going to talk to him about the impact of AI on business growth?

Welcome, Sumanth.

Sumanth Vakada: Hi Sahil.

Sahil Brij Malhotra: So Sumanth, last time we were having a discussion around how analytics and AI are important and how they have changed the way businesses are being done at least over the last decade or so. Given that background what has happened is that AI and analytics efforts implementation specifically all of a sudden have been attracting a lot of investments in that particular space. While AI and analytics efforts have been attracting a lot of investments in the space, why are companies still struggling to implement AI in meaningful ways? So could you help us understand that a little?

Sumanth Vakada: Sure Sahil, that’s a good question. So let us look at the problem that you’re talking about. So the problem that you’re talking about is typically called a scaling AI challenge and what it typically means is companies usually don’t have a problem experimenting with AI what they usually do is build a small proof of concept using an example AI use case and achieving success with the proof of concept is not that difficult. But where I’ve seen companies falter and usually, a test failure is with taking that proof of concept and scaling it to a production level implementation and then making that AI model run in real-time with the data that’s flowing in from all of the different data sources. So that is where I’ve seen many companies fail or attempt and fail with implementing that properly. In fact, the success rate with implementing AI at the production level or scaling AI successfully is in the low double digits or in the single digits. And the companies that are usually having success in this are the ones that are pouring a lot of money into making this a success at a very high cost. So it is a problem that needs to be solved. Before I tell you what are the factors that are causing this kind of failure, let me ask you a question. What do you think is the primary objective of a company that is operating in the market today?

Sahil Brij Malhotra: Well, Sumanth for any organization that is operating in its competitive landscape, I think the biggest goal or objective for such a company is to generate as much high profitability as possible.

Sumanth Vakada: Good. I am extinguishing that a bit further beyond profitability. There are companies that are focused on achieving certain missions whether they are for non-profit or for environmental reasons for those kinds of companies, it is also important to achieve those kinds of objectives. And the reason I brought up that is this kind of companies, whether they are they’re trying to achieve a higher degree of profitability or they’re trying to achieve some objective or some mission that is linked to the company there they are the ones that are depending on implementing AI to give them the insight from the data that they have to make sure that they take the next best action possible. So that means that the companies that are depending on AI to give them the right insight have to have access to the right kind of data, the most relevant data, the most contextual data, and the most up-to-date data, that can help them take the next best action. So looking at traditional reporting it’s no longer useful to just say that you have achieved a million dollars in sales in the last month. What is more important and what is more relevant to the businesses is telling the business how they can achieve the next million dollars in sales and how they can maximize the profit in that million dollars of sales. That is what companies are looking for. If you paid attention here to what I said, data is the key central asset here. So making sure that you have the right kind of data, the most contextual data, and the most up-to-date data, is very important, and not having that is the number one reason for companies to fail when they are implementing any kind of AI project. And in order to overcome that, what we advise companies is to have a data strategy. Where you are looking at all of your different data sources, identifying and inventorying all the data points that you have, making sure of the structure of data, and identifying the relationship from the data to another data point. And in the end, what you’re essentially doing is building out a framework of data points from different sources and I would call it building or manufacturing a data fabric, of all your data inventory that you have. And I call it a fabric because just like how you have a fabric where you pull on one end you feel it on the other end similarly when you have a properly structured data fabric, you can see how one data point impacts another data point from a different data source. So, having that kind of data fabric is very important and the process to get there is what we call a data strategy.

So the number one factor for success is having a data strategy and not having that is also the number one reason for a failure in implementing AI. The second most important factor of having a data fabric using a data strategy is having something that we call an intelligence strategy. We have talked about this a lot. And, we have even published an ebook on this. What we mean by an intelligence strategy is making sure that you are asking the right questions from the data that you have. That is one of the most important steps. And it may seem simple but it’s an incremental next step from having a data strategy. Unless you ask the right questions from your data, you will not be able to identify the right outcomes that you can derive from the data that you have, and from the questions that you asked. I’ve seen many companies do this in a backward way where they identify an AI outcome that they would like to achieve and then they try to retrofit that AI outcome to the data that they have. And usually, that’s the recipe for disaster. Because you are trying to create an artificial outcome from data that you don’t necessarily have and that is never successful. And I have seen that happen quite a few times. So having this incremental process of building a data fabric and having an intelligence strategy to ask the right questions allow you to be on the right path to implementing AI on whatever scale it may be possible. You may not be able to implement AI to the fullest degree that you would like to but at least you will be implementing a successful AI model in this process. And the by-product or the side effect of following this process is that if you identify that you don’t have certain data points then you go back to your data strategy and then add on those data points or identify those data sources where you can capture those data points and then you recreate this process. So those two are the most important things. Now the third most important factor is making sure that you have a stakeholder that oversees the entire process. And the responsibility of the stakeholder is overseeing the process, overseeing the milestones, identifying the resources that you need, identifying the overall objective that you want to meet and manage the team, or working with the team to get there. So that is an extremely important step on top of having a data strategy and an intelligent strategy. And then another responsibility of this key stakeholder is governance, making sure that you have the right governance structure to make sure that you are using data in the most appropriate way to make sure that you’re driving the right outcomes.

Sahil Brij Malhotra: Right. very good points mentioned by you Sumanth. So basically for any company which is looking forward to implementing AI and analytics efforts seamlessly, they must look at firstly their own data inventory which will help them weave out a data strategy. Second, they use that particular data inventory in data points from that data inventory they must ask intelligent questions and relevant questions which actually will help them align with and achieve the objectives they are out to achieve. Thirdly, you mentioned the presence of a key stakeholder who is going to overlook this entire process of seamless implementation of AI and analytics efforts exactly. Now given that background could you know there are a couple of questions that kind of come to mind directly, first when we look at this key stakeholder does this key stakeholder need to be somebody internal from the organization or somebody who needs to be hired externally? Second, does this particular key stakeholder need to have a previous AI analytics implementation exposure or not? Thirdly when we look at this particular key stakeholder in terms of implementing this entire AI process don’t you think that somebody from the c-suite or somebody from the top management or senior management can double up in terms of responsibility for implementing AI and analytics in addition to what they are doing? So could you help us understand this a little?

Sumanth Vakada: Sure Sahil. Another good question. I will answer your questions. So I will start by answering your first question. Whether this person is an internal person or an external person is purely the choice of the company, it depends on the comfort level that the company has with the person and how committed this person is to drive the success of the company. So I think I would leave that up to the company to make that decision. The second factor is whether this person can have or should have a background in implementing analytics and AI initiatives or projects, I would say that is secondary to the fact that the person or the stakeholder needs to have a very good understanding of the market and the business that it’s operating in. And, the third factor is what you ask for whether the person is should be part of the c-suite? They can be, but what I would say is that the responsibility of this person or the stakeholder should be primarily focused on AI and thus, AI should not be a side project that should not be a secondary project for the stakeholder. It has to be the only, if not the most important focus of implementation for the stakeholder. I say that is because implementing AI is extremely complex, so that is the reason why many companies start with a proof of concept. And proof of concepts is usually expensive. But what I’ve seen happen is when a company is trying to implement AI and they start with the proof of concept. They start with immediately diluting the overall effectiveness of the AI, by picking a proof of concept, which is a narrowly defined scope compared to the full-scale implementation, and then when that project changes hands from one function to another function what you receive or what you achieve at the end of the day is a shadow of what you wanted to achieve. And that usually leaves people with very little motivation to go and then invest in a full-scale AI project. So that is why having a stakeholder that is completely committed to the success of AI and maintaining the focus, motivation, and determination to implement AI successfully from start to the finish is very important. And this will have to be their primary focus. So more than any other thing I would say that the most important factor that the stakeholder needs to believe is that they are committed to the success of AI. They need to have a very good understanding of the market the business is operating in, and they should be able to work with the entire organization and any function within the organization very seamlessly, very effectively. So that they are maintaining the right levels of engagement and motivation to complete the project successfully. So, I believe that if these three factors are in place, having a very good data strategy, having a very good intelligence strategy, that is managed by the stakeholder, and making sure that they’re following this process correctly, I would say that many more companies would be tasting success with implementing AI.

Sahil Brij Malhotra: Right, some very good points mentioned by you, and thank you for enlightening us. Specifically, on the aspect of the stakeholder who’s going to be the central pivot around which the entire AI and implementation efforts will be seamlessly integrated into the organizational goals. So in a nutshell any company which is looking at implementing AI and analytics efforts seamlessly must have the right data strategy in place, the right intelligence strategy in place, and the presence of a key stakeholder who is not only accorded with the responsibility to only look at AI implementation successfully and meaningfully but also this individual – a person needs to have as rightly said by Zara’s senior Ortega, five fingers on the market and five fingers on the customer in order to understand how important it is to look at the implementation of this effort successfully. And just to top it up Sumanth, I would say as you rightly mentioned that data is the central key as a set for any organization. Any organization must always kind of categorize data as a separate asset category like we have current assets in any organization. So I think there must be a separate category of assets that kind of looks only at data this will not only help organizations in generating meaningful insights but also help the organizations in generating incremental revenue generation opportunities and higher profitability in the case of for-profit organizations and also in the case of not-for-profit organizations or mission-driven organizations they would be closer to what they want to achieve in terms of their mission, In the case of not for profits so really thank you for the insights on this Sumanth, thank you for excellent points. Thank you.

Thank you, Sumanth for your insightful thoughts on the impact of AI on business growth. We just heard from Sumanth Vakada, founder and CEO of Qualetics Data Machines. We will be back with more thoughts from Sumanth and our other team members soon.

Thank you!