How agentic AI is reshaping banking in South Africa

Rewind only a few years and enormous language fashions and generative AI have been barely on the general public radar. But the know-how has already advanced into its subsequent iteration: agentic AI, a brand new breed of methods which might be semi- or absolutely autonomous and in a position to cause and act on their very own. And adoption has skyrocketed, notably over the previous 12 months.

In response to a latest PwC survey, 79% of senior executives globally say AI brokers are already being adopted of their organisations in a single type or one other, which is a placing stage of uptake for such a nascent know-how.

One would possibly assume that agentic AI has not but reached South African shores, however it’s already being applied at scale, notably within the monetary companies sector, the place early deployments are delivering measurable features in effectivity and productiveness.

For purchasers, it means sooner service, extra personalised interactions, faster decision of on a regular basis banking wants and fewer time spent navigating routine processes.

Take the shopper acquisition stage for instance. Onboarding a small or medium-sized enterprise has historically taken a number of hours unfold throughout a number of days, with relationship managers amassing paperwork, verifying enterprise registration, conducting “know your buyer” (KYC) checks, assessing credit score danger and configuring merchandise throughout completely different methods. That stage of handbook work drives up the price of buying every buyer, which in flip limits how broadly banks can serve the market, usually forcing them to prioritise bigger, higher-value purchasers over smaller companies.

Enterprise agentic AI

Now think about compressing that course of into 20 minutes by means of automation, utilizing instruments akin to biometric identification verification, automated CIPC lookups, real-time credit score bureau integration and prompt KYC orchestration. Out of the blue the economics change and banks can serve segments that have been beforehand not viable at scale, increasing their addressable market and opening the door to far broader monetary inclusion. For the SME proprietor, it means having the ability to open a enterprise banking account in minutes quite than days, with far much less time spent navigating paperwork and administrative processes.

Many banks throughout South Africa have already begun investing in these applied sciences, notably in business-critical operations, and their software throughout different areas of the banking worth chain is predicted to broaden significantly by means of the remainder of 2026.

Learn: The actual cause Absa wrote off R2.4-billion in software program

In September final 12 months, Absa, for instance, partnered with cloud-based CRM platform Salesforce to carry its enterprise agentic AI resolution, Agentforce, into Africa for the primary time, a growth that carries essential implications for the banking business.

As a part of this, three autonomous AI brokers are being examined throughout the financial institution:

• The primary is a copilot designed to help relationship managers. Traditionally, a lot of their time was spent getting ready for conferences quite than partaking prospects, with as a lot as 75% of the workday going into reviewing buyer knowledge throughout a number of methods, analysing latest transactions, figuring out potential alternatives and compiling engagement notes. Solely about 25% of their time was spent really interacting with prospects. The AI agent now automates a lot of that preparation by producing pre-meeting briefs, pulling collectively buyer data from a number of methods, highlighting related transaction patterns, and capturing notes after engagements. This has successfully flipped that ratio, with administrative time lowered by between 75% and 90%, permitting relationship managers to spend way more of their time centered on prospects.
• One other agent focuses on buyer enquiries. It operates as a multilingual help assistant able to responding to frequent banking questions throughout 11 languages, together with isiZulu, Sesotho, and isiXhosa. Early outcomes present that round 40% of customer support queries are actually resolved with out human intervention, which means 4 in 10 interactions will be dealt with immediately whereas sustaining response high quality. In lots of circumstances, responses that when took prospects as much as half-hour to obtain are actually delivered instantly, with success charges of near 99%.
• The third agent focuses on monitoring and resolving technical points throughout methods. Working across the clock, it could possibly detect and deal with many issues routinely, lowering the time it takes to revive companies and permitting groups to focus much less on operational troubleshooting and extra on supporting prospects. Since its introduction, the agent has already dealt with greater than 6 400 inner help queries with a hit price of about 96%, which means the overwhelming majority of points are resolved without having to escalate to technical help groups.

Faster entry

These examples provide a glimpse of how agentic AI is starting to vary the best way banks function and, extra importantly, what prospects can anticipate within the close to future.

It’s possible that extra of this know-how might be used to broaden digital onboarding, permitting prospects to open accounts or entry companies way more shortly and with much less paperwork.

Banks may additionally start utilizing AI to evaluate creditworthiness in new methods, analysing transaction patterns and money move behaviour quite than relying solely on conventional steadiness sheet data. For small companies particularly, this might imply faster entry to financing and extra proactive gives when their monetary exercise reveals they can help extra credit score.

On the similar time, many routine banking interactions are more likely to transfer into digital self-service channels, permitting prospects to resolve easy requests immediately whereas nonetheless accessing human help when it’s genuinely wanted.

Agentic AI remains to be in its early phases, however its route of journey is already changing into clear.

As these methods mature and banks turn into extra comfy deploying them in on a regular basis operations, prospects are more likely to expertise banking that’s sooner, extra responsive and more and more personalised. A lot of this alteration could occur within the background, however over time it’s going to remodel how individuals work together with their banks.

• The creator, Lindelani Ramukumba, is CIO and interim chief digital officer of Absa Enterprise Banking