In the race to embrace AI, organisations often overlook a critical stage that can turn successful AI adoption into a costly disappointment. According to Jonathan Oaker, founder and CEO of CloudZA, most South African enterprises are still running core business logic on legacy systems. This infrastructure is siloed, brittle and incapable of processing data at the speed modern AI requires.
This is where businesses regularly get stuck, he says. “They recognise that their legacy environment is holding them back, but they're unsure how to move forward. In response, many simply lift-and-shift legacy workloads into the cloud, believing the hard work is done and that they’ll soon be able to unlock the full value of AI.” But just because a business has migrated legacy applications to the cloud doesn’t mean that they're AI-ready. “In fact, a lift-and-shift simply relocates the constraints the organisation was trying to address.”
Is this why your AI pilot failed?
Let’s imagine a contact centre pilots an AI solution to automate quality assurance. In the past, supervisors could only review a small sample of calls each month because the process was so time-consuming. But when using AI to analyse call recordings and transcripts, it’s possible to assess every customer interaction. It's an ideal proof of concept because the data is consistent and predictable. But when the organisation looks to extend AI beyond quality assurance, things fall apart. Instead of analysing one clean, structured data source, the AI must now work across a data environment that is significantly more complex.
As part of this more complex environment, let's say there's a traditional SQL-style database. This database would need an index to make it queryable. “If it's a big dataset, you’d also need some form of partitioning, which is very difficult to achieve on legacy data structures,” says Oaker. Without modernising these systems and establishing consistent, governed data, a pilot that delivered impressive results in one department can struggle to scale.
Modernisation matters
Modernisation is about laying the right foundations for AI success. When it comes to modernisation, organisations have one of two options. They can either take a traditional database and put it on a modern layer, which provides a serverless environment. This approach bridges the gap between legacy and modern environments, essentially creating an advanced legacy environment that AI can consume.
The alternative approach is to go completely cloud native, he adds. “I believe that this is going to be the biggest change many businesses make over the next couple of years. Here you take the source data out of a transactional database and move it to the cloud, which makes it much easier for AI to reference it directly. The approach also provides governance and other privacy controls, which are very difficult to apply in a traditional data storage environment.”
As a partner, CloudZA will work with organisations to understand their legacy systems and schemas and then come up with smart ways to modernise these environments. Working across the entire cycle – from legacy and migration through to modernisation and automation/AI – to ensure the business gets value from start to finish.
CloudZA also runs workshops with customers, spending several days with the business and working closely with different teams to determine what will drive the most success. Through this process, CloudZA generates a ROI report that outlines what the customer is currently doing and how much it’s costing, and what this scenario could look like within three to six months with new processes in place.
Want to assess your AI readiness? Visit the CloudZA website to do a quick two-minute assessment.
Share
