Johannesburg’s AI conversation seems to be shedding its costume jewelry at last. Zensar’s invitation-only CIS Converge 2026 forum, held on November 20, put the attention where it belongs: not on breathless demos, but on whether companies can turn AI into sturdier operations, better customer experiences, and something resembling a measurable return.
That is a healthier discussion than the usual parade of grand promises. Plenty of executives want AI plastered across the annual strategy deck. Fewer relish the unglamorous work underneath it: cloud bills, inconsistent data, fragile systems, identity controls, and the plain old discipline of running technology reliably. Yet that groundwork decides whether an AI project becomes useful or merely expensive theater.
Infrastructure Is the Part Nobody Can Skip
Zensar’s central message was blunt enough: AI results are only as good as the infrastructure, security, and operational model supporting them. Jitendra Nandwani, the company’s SVP and Head of CIS, made that point directly. It is not revolutionary, but it bears repeating because the market has spent years acting as if buying models somehow replaces the need to manage systems.
The firm highlighted Autonomous Operations, FinOps, multi-cloud operations, and hyperconverged infrastructure as tools for handling that mess. Those are not flashy phrases, and thank goodness for that. FinOps in particular matters because cloud spending can turn into a leaking roof if nobody owns the meter. AI workloads may amplify that problem, especially when enterprises spread applications across multiple cloud providers without clear cost accountability.
Security Gets Harder When Machines Gain Identities
The most useful thread from the gathering may have been identity security. Saviynt’s Alex Paterson pointed to the rise of AI agents and non-human identities, arguing that organizations need fresh ways to secure and govern them. Correct. A business can control employee access imperfectly and still survive. Let a growing population of automated agents access systems, move data, trigger workflows, and make recommendations without sound controls, and the risk profile changes in a hurry.
- Who authorizes an AI agent to access sensitive information?
- How is its activity monitored and audited?
- What happens when a model, cloud platform, or vendor connection fails?
- Can management show that automation reduced costs or improved service, rather than simply shifting risk?
Those questions are not a buzzkill. They are the difference between responsible adoption and a future cleanup job that nobody wants on their hands. AWS, Microsoft, ServiceNow, and Saviynt were among the partners represented, which also underlines a basic reality: large enterprise AI deployments will remain ecosystems, not one-vendor miracles.
African Enterprises Want Proof, Not Posters
Kaushik Chatterjee, Zensar’s SVP and Head of Africa, said enterprises across the continent are increasingly focused on practical AI uses with tangible business outcomes. That sounds believable, and frankly it is where the conversation ought to be. The notes do not provide customer contracts, revenue targets, deployment figures, or quantified savings, so investors should resist reading this event as proof that demand has already translated into financial results.
Still, the direction is worth watching. Banking, insurance, healthcare, manufacturing, and telecommunications all face real pressure to improve service quality while controlling technology costs. A service provider that can help clients modernize infrastructure, govern identities, and operate complex cloud estates has a credible place in that spending cycle. The catch is execution. Consulting and transformation work can be sticky, but it is also crowded, labor-intensive, and vulnerable to customers delaying projects when budgets get tight.
Speed without reliability is a shortcut to trouble; reliability without speed eventually loses the customer.
Shaun Pollock’s keynote framed that balance through leadership and sustained performance. It is a familiar sports analogy, sure, but it fits this subject better than most. Enterprises should move quickly enough to learn where AI helps, while refusing to gamble core operations on immature controls. There is no medal for being first to deploy a system that creates a security headache.
What Deserves Attention After the Event
Zensar, part of the USD 5.2 billion RPG Group, is positioning itself around experience, engineering, and engagement technology across several major industries. This forum reinforces that positioning, but a forum is still a forum. The evidence that matters later will be client wins, repeat work, project margins, concrete cloud-cost improvements, and credible examples of AI deployments surviving real production pressure.
For now, CIS Converge 2026 gets one important thing right: the AI gold rush is entering its less glamorous phase. The winners may not be the loudest vendors. They will be the ones helping customers keep systems secure, costs visible, and operations dependable while the technology keeps changing under their feet.