5 ingredients for enterprise success in AI

AI has recently become the latest buzz amongst industry sectors. Honestly, you’d be hard pressed to find a business that hasn’t started talking about it. But, before we start implementing AI in our own enterprises, it’s important to understand not only the benefits AI can provide, but also the foundations required for it to be successful.

Some quick online research will reveal that there are innumerable schools of thought and ideology on AI and its use. And all that ‘noise’ is great because it generates interest in this amazing technology. Importantly, one thing remains clear: businesses must start their AI journey with a concrete business problem if they truly wish to reap AI’s full benefits.

Let’s take an age-old problem as an example - effective ways to deal with back orders! The ultimate supply chain conundrum. With every order missed, an unhappy customer and a negative revenue impact surface. Furthermore, with the advent of the digital world and supply chains’ accelerated pace, the need for a solution has become even greater. This became evident during the ongoing pandemic when consumerism was disrupted across the world. A question that naturally comes to mind is whether there’s a way to efficiently predict which items will require a back order, and automatically re-order said items to increase optimisation?

While defining the business problem is the first step forward, an essential ingredient for AI to work is data. Let’s revisit our supply chain example. Does the enterprise even have historic data of the items going out of stock? Only if the data exists and is reliable in nature, could there be a business case to implement a machine learning (ML) model. Historic data is foundational to the creation of relevant ML models which make an ecosystem artificially intelligent. It’s a valuable gold mine to any enterprise.

According to IDC, by 2025, the volume of data generated worldwide will reach 175 zettabytes, an astounding 430% increase from the 33 zettabytes produced in 2018 (for reference, 1 zettabyte is equivalent to a trillion gigabytes). It is critical for any business committed to data-driven decision-making to unearth the patterns and insights hidden in their vast amount of data. AI provides enterprises with the competitive edge to capitalise on their data and charge ahead of the competition.

It is important to flag that prediction alone will not help the enterprise. How we use the prediction in the overall business layout will dictate the efficiency of implementing AI. For example, do we kick start a replenishment order just in time? Do we remove the stock entirely, or do we wait for a threshold to trigger? Each option has its own pros & cons and will yield different business outcomes dependant on the priority focuses or KPIs of the business. Nevertheless, AI will allow the business greater and more accurate insights, supporting the overall decision.

So, is it all hunky dory?

It is a myth that AI can solve all business problems. Enterprises need to perform due diligence on identifying the business use cases that fall under the remit of AI. Another important aspect is understanding that enterprise data is the fuel of AI.  To ensure an efficient and error free model, the underlying data should be clean and free from bias. In other words, AI is taking the phrase ‘garbage in, garbage out’ to a whole new level.  Finally, the outcome of any model should be explainable in nature. No matter what state of the art model is used, your enterprise will not appreciate the efforts if they are unable to comprehend the results produced by AI.

So, what does it take to kick start the AI journey?

Recent studies on AI adoption show that AI deployments are rising. Gartner predicts that the average number of AI projects per company in 2022 will grow to 35, a 250% increase from 10 in 2020. However, different organisations are at different levels of maturity in the AI journey. There is no single defined path that will be a silver bullet for all. Despite this variation, some of the key focus areas for kick starting any AI journey includes:

• Building the data estate –data is key to the success of all AI programs so enterprises must ensure a strong data framework
• Identifying the key business challenges – review current business problems and establish ready to implement pilot uses cases
• Shortlisting the right team – the AI team should be multi-skilled and cross functional with special emphasis on domain knowledge
• Building the MVP – a key mantra of success is to start small, understand the gaps, course correct and then scale
• Planning for scale – while small experiments are a great place to start, enterprises need to plan to scale their AI to yield a maximum benefit.

To turn AI into a reality, enterprises need to ensure they have the right mindset for successful AI adoption, the right skills in their teams and an AI execution strategy. Technically skilled individuals with rich domain knowledge are important but they are not the only skill required. Players who know the enterprise IT landscape and business cycles are equally valuable. While small benefits are exciting, it is critical to have the big business goals in mind to have a sustainable AI journey.