Rachit Tayal, associate VP and country head at HCLTech, shares real-world insights on moving from AI concepts to tangible business value in South African companies.
So, you’ve got your organisation’s five-year roadmap, now what? The strategy is in place, the board is aligned and AI is high on the agenda. But turning a vision into value is where the real complexity begins, and something the CIO often grapples with. What does it take to make AI actually work?
While many South African organisations rush to implement AI solutions, few have clear frameworks to measure success or practical roadmaps to guide implementation. As CIOs face pressure to show returns on technology investments, distinguishing between AI hype and actual business value becomes critical.
“AI is one of the topmost things CIOs think about, talk about and worry about. It’s changing the world we live in and the work we do at all levels – not just in a specific scenario,” says Rachit.
At HCLTech, a leading information technology consulting company, Rachit works with CIOs to map out realistic journeys from their current positions to their desired destinations. This practical approach avoids the common pitfall of implementing technology without clear business objectives.
The three broad AI adoption categories
According to Rachit, the South African market can be grouped into three broad AI adoption categories (or buckets) that companies will look to adopt:
1. Productivity-focused AI: Productivity tools, such as copilots, are where he sees the most traction. When writing a Microsoft Word document, for example, one can use a copilot to edit drafts. Or it can be used to quickly draft and respond to emails. In South Africa, most organisations have adopted productivity-focused AI well.
2. Embedded AI: The second bucket includes the AI capabilities embedded in core enterprise applications, like Salesforce, for example, or a company’s finance and budgeting system – basically all commercially off-the-shelf software that caters not to the masses but to specialists who actually use it.
Rachit notes that adoption is slowly picking up here, but it’s not where it could be. “I think there is still a long way to go before these things are fully adopted,” he says. “In the conversations I have with business leaders, they are trying to educate their teams on how to better use AI in their day-to-day operations and identify areas of transformation where it could meaningfully change the way they do business.”
3. Custom AI solutions: The final bucket is comprised of generative language models that businesses are using to create custom AI solutions. “In such scenarios, I think the adoption depends on the quality of the use case,” Rachit explains. “When you are trying to solve a seriously meaningful business problem, you will see these use cases get adopted and deployed more.”
He believes there’s still considerable potential for growth across all three buckets.
The three measures of AI success
Instead of vague promises, Rachit recommends judging AI implementations against three specific KPIs:
1. Productivity and speed: AI solutions should solve unnecessary bottlenecks and barriers by doing things in a more efficient manner. “It should enhance the overall productivity of all the people who are engaged in that specific process or problem,” he says.
2. Quality: For people to actually use AI systems, the quality must be spot on. As such, notes Rachit, reduced error rates, enhanced decision-making abilities and accurate outputs are very important. “You can’t have hallucinations – so if you don’t have guardrails, you might have bigger problems.”
3. Business value: The use case must deliver some differentiating value to the business process. Speed, accuracy, new insights resulting in new revenue opportunities and process simplification are great examples of value differentiation. “Business value delivered is a key measure of ROI, whether it’s improving the customer experience, launching a new service or streamlining your internal operations,” adds Rachit.
Building responsible systems
While the conversation around making AI work often focuses on capabilities, one mustn’t overlook responsibilities. Rachit emphasises again that governance must be part of the initial design, not an afterthought.
“We start every engagement with a dedicated team focused on responsible AI implementation,” he says. The framework for responsible AI includes three important elements:
- Process and technology guardrails built into the foundation
- Risk assessments tailored to each company’s and project’s unique challenges
- Broader societal considerations to ensure positive community impact
“We should do these engagements with a mindset of uplifting society,” Rachit encourages, noting HCLTech's commitment to developing local talent in South Africa. “We encourage our clients to build these capabilities inhouse, rather than remaining dependent on external expertise.”
Getting South Africa tech-ready
A local financial services company worked with HCLTech to improve revenue and efficiency, using tech to transform how they handle complex policy documents in the sales and customer support teams. “The cost of acquiring a customer is very high in the financial services industry, so they needed to make both sales and support processes much more efficient,” Rachit explains.
The solution was an AI system that analyses lengthy legal and policy documents, making them easier for support agents to navigate while speaking to clients on the phone. “We even built it out further to help the customer conduct internal reviews based on actual customer interactions,” he shares. This transformed what had been a frustrating manual process into happier customers and valuable insights for the teams designing insurance products.
But technical implementation is only half the equation. The human element determines whether AI projects succeed or fail. “We have a very comprehensive edtech offering, which we are encouraging CIOs to use to build an AI-ready workforce,” Rachit says. This platform offers accessible training, allowing individuals to “do their own learning, take exams, certifications and come out much more enlightened and AI-ready”.
HCLTech also conducts AI immersion days and workshops with senior leadership to take them through tech trends around the world and explore what the current imperatives are, and how they can embrace them. Rachit points out that these training sessions have proven valuable in ensuring that “AI is not only a CIO problem”.
For South African CIOs planning their next steps and refining their five-year strategies, Rachit encourages them to start where they are and focus on what their people need. “AI should always serve a meaningful purpose, rather than existing solely as a technological novelty. Those who approach AI with clear objectives and practical measures will stand apart in an increasingly competitive landscape – not because they implemented AI first, but because they implemented it effectively,” he concludes.
HCLTech operates in 60 countries and has been active in South Africa for over 14 years. The company holds level one B-BBEE contributor status with 75 percent black females and has been ranked among the top two employers in South Africa for the past three years.
