Leveraging AI technology for
business success

An approach to AI based value creation

After a decade-long promise of AI creating value for companies, it has finally matured to the point of delivering viable solutions. Although not perfect, the natural language model ChatGPT 3 has demonstrated how simple prompts can generate outputs that increase productivity, automate tasks, and produce creative assets.

This has sparked a renewed interest in AI technology as a whole, with companies worldwide striving to comprehend its potential. Additionally, they are grappling with the reality of a potentially disruptive technology they do not know how to leverage.

Responding to AI technology

Two approaches have emerged in responding to AI technology: wait-and-see, and test and learn. Many companies are opting for the first approach, waiting for evidence of AI’s promises.

A smaller number of pioneering companies are diving in and conducting experiments. They are identifying use cases and designing minimum viable products (MVPs) based on AI technology. They either adopt promising solutions or iterate with new use cases. McKinsey surveyed some of these pioneers and found a group whose EBIT increased by more than 20% after implementing AI solution.

In this frenzied environment where new AI products are released frequently, opinion is split on the usefulness of the technology and expertise in the area is in short supply

Based on conversations and sentiment, many companies in South Africa are adopting the wait-and-see approach. This may be prudent, but it could mean missing out on opportunities to strengthen competitive advantage and increase profits. In this frenzied environment where new AI products are released frequently, opinion is split on the usefulness of the technology and expertise in the area is in short supply; it is possible that the wait-and-see approach is a sign of a deeper issue – companies may not know where to begin when engaging with the technology.

Asking the right questions

Questions arise, particularly around selecting the “correct” tools and applications. While it is reasonable to ask these questions, they are a step too far from where businesses ought to be in their thinking. Better questions to ask are narrower and more practical. They help taper possibilities and bring focus, reducing the paralysis companies face. Here’s a list of possible questions to ask as a business leader:

  • What problems does my business face that can be addressed through automation?
  • Where are my employees engaged in repetitive low-value tasks?
  • Does my business regularly deal with large, unorganised datasets that are challenging to analyse?
  • As a large corporation, can this technology help organise the vast amount of data we collect?
  • Are there any successful areas where companies similar to mine have implemented AI?
  • What is the safest and most cost-effective way for me to personally test the technology?

Fortunately, businesses do not have to start from scratch, thanks to the companies that have already tested AI technology.

Several use cases have been successfully addressed with AI. While some use cases are specific and not easily generalised, a number of them apply to all companies that rely on technology in their operations.

The following areas are the most promising for businesses to explore when considering AI solutions within their operations:

  • Automation: AI can be used to automate repetitive and mundane tasks, such as data entry, report generation, and basic customer inquiries. This improves efficiency, reduces errors, and frees up employees’ time for more complex and strategic work.

  • Security: Security measures can be enhanced by artificial intelligence by analysing patterns and detecting anomalies in real-time. It can be used for intrusion detection, threat monitoring, and identifying potential vulnerabilities, helping businesses protect their digital assets and prevent unauthorised access.

  • Fraud Prevention: The technology can analyse large volumes of data to identify fraudulent patterns, transactions, or behaviours. By leveraging machine learning algorithms, businesses can build models that continuously learn and adapt to emerging fraud tactics, improving detection accuracy and reducing financial losses.

  • Recommendation Systems: AI-powered recommendation systems can analyse customer behaviour, preferences, and historical data to provide personalised product or content recommendations. This enhances the customer experience, increases engagement, and boosts sales by suggesting relevant offerings to individual users.

  • Customer Service: AI can be used to enhance customer service through chatbots and virtual assistants. These AI-powered tools can provide instant responses to customer inquiries, offer self-service options, and handle routine support requests, improving response times and customer satisfaction.

  • Marketing: Audience segmentation, predictive analytics, and campaign optimisation are various marketing activities AI can assist with. By analysing large datasets, AI can help identify target audiences, personalise marketing messages, and optimize advertising spend for better campaign performance.

  • Trading: Intelligent algorithms can analyse vast amounts of market data and perform real-time analysis to identify trends, patterns, and investment opportunities. AI-powered trading systems can automate trading strategies, execute trades faster, and optimise portfolio management for improved financial outcomes.

  • Safety: Workplace safety can be improved by its application to monitor and detect potential hazards or risks. It can analyse sensor data, video feeds, and environmental factors to identify safety issues, alert personnel to potential dangers, and facilitate proactive measures to prevent accident.

Successful implementations of AI technology

Several leading companies have successfully used AI technology to solve complex problems and achieve profitability.

One such example is Carmax, the largest retailer of used cars in America. When their customers conduct pre-purchase research, they partially rely on information from other customers, such as car model reviews. The challenge is there could be thousands of reviews per car model. To address this challenge, Carmax utilised OpenAI’s cloud-based API and later Microsoft Azure Open AI Service to create summaries of 5 000 car pages. The process took just a few months to complete; they estimate it would’ve taken 11 years using their current manual processes.

We conducted our own experiments, utilising chatGPT for specific business problems in different industries. Overall, they produced promising results when the unique context and specific data for the use case was thoroughly interrogated.

One scenario involved using chat GPT to recommend risk-mitigating controls based on the risk type and department. When the results were examined by experts in the field, it was noted that generally, while there were gaps, the tool produced adequate results and demonstrated potential to quickly identify best practice industry controls to proactively manage risk.

Limitations of AI

While AI offers numerous benefits, it is important to note that it is not a silver bullet solution. The aforementioned experiments have revealed certain limitations of the technology, including genericness, uncertainty, and safety and reliability. Companies should keep these in mind as they begin to experiment.

  • Genericness: solutions are not designed for specific domains but rather as one-size-fits-all solutions. When untrained with context specific information, they may lack depth and specificity when applied to specific problems.

  • Uncertainty: The viability of intelligent tools is clouded by unanswered questions. This is the result of many of them still being in the development stage. It remains unclear how they will integrate with existing systems and processes, the security risks they may pose, and how to address the unknown unknowns surrounding them.

  • Safety & Reliability: AI systems function as black boxes, where we only observe inputs and outputs. The internal processes that drive the models and algorithms are not well understood. This raises concerns that these systems may expose businesses to risks they do not understand and cannot mitigate.


Artificial Intelligence is finally beginning to deliver on its decade-old promise to revolutionise business. We have reached a point where businesses can cautiously design their own experiments based on specific, well-defined use cases. While some experiments may fail, those that succeed are likely to generate significant benefits, leading to increased efficiency, higher revenues, and an enhanced competitive edge.