The era of adopting artificial intelligence simply because it is the “must-do cool thing” is officially over.
As African businesses look toward 2026, the focus has shifted entirely from broad experimentation to demonstrable return on investment (ROI).
According to Cliff de Wit, chief innovation officer at ADG, the market is entering a phase of discernment where business value takes precedence over novelty.
Moving from experimentation to measurement
De Wit says that the initial wave of AI adoption saw many organisations investing large sums without a clear understanding of the return. That era of experimentation is rapidly changing. While customers know they need to implement AI, they are becoming increasingly specific about the why.
To unlock value, businesses must find the “sweet spot”: high-impact initiatives that are not prohibitively expensive to start, says De Wit. “While there is no definitive global standard for measuring AI ROI yet, it cannot be treated as “magic and fairy dust”.
Businesses must measure the cost and complexity equation against the impact to decide if a project provides value before sinking capital into it.”
Employee productivity as the first frontier
While many companies look immediately to customer-facing chatbots, De Wit argues that the most tangible ROI often comes from servicing employees first. Most organisations must balance internal capabilities with customer services. Starting with internal productivity offers results that are easier to measure.
“Starting by improving employee productivity is a good way to get ROI,” explains De Wit. “For example, using an AI bot to assist a legal team can reduce work that would usually take weeks into minutes, saving significant time and fees.”
The human factor: training and change management
The South African market is reasonably well-tooled to make use of AI. The widespread use of consumer tools like ChatGPT and Gemini has readied the workforce, meaning many employees have already experienced the potential of AI in their personal lives. The gap, however, lies in translating this to the work world.
This is also creating a new layer of risk. Employees are using personal AI assistants to handle work tasks, and unintentionally exposing confidential data to public models. Businesses need to manage this enthusiasm and translate it into a secure, managed, work environment.
De Wit says that while companies provide access to tools, they are often under-invested in the training and change management required to get a productivity benefit. “We find organisations that train their staff tend to understand how the tooling works. Just to start with – once you create that curiosity, people start learning themselves,” he says.
This training is critical to overcoming natural resistance. There is still a stigma attached to AI, often driven by a fear of job replacement. De Wit notes that in practice, jobs are rarely replaced; instead, productive employees simply produce better quality output. “Using AI without human oversight is full of gaps, even with today’s tech,” he adds.
2026 trends: specific context over generic data
Looking toward 2026, the major trend driving ROI will be a shift from generic tools to specific, intelligent agents. Organisations are realising that moving beyond standard vendor tooling to build their own agents takes them to the next level.
This involves training agents on an organisation’s own content. This provides specific, contextual insight rather than generic data scraped from the wider internet.
“Organisations using generic AI tools will start adopting and building tools specific to their content, context, and industry to ramp up capability,” predicts De Wit.
As the technology matures, tools are becoming much more capable in specific areas of expertise, such as creative asset generation. The organisations that will succeed in the coming financial year are those that become clearer on exactly what they want out of their AI use cases.