Rewind a few years and big language models and generative artificial intelligence were barely on the public radar, yet the technology has already evolved into its next iteration: agentic AI, a new breed of systems that are semi- or fully autonomous and capable of reasoning and acting on their own. And adoption rates have skyrocketed, especially in the last year.
According to a recent PwC survey79% of senior executives globally say that AI agents are already being adopted in some form in their organizations, which is a remarkable level for such a budding technology. One might assume that agentic AI has not yet reached South African shores, but it is already being implemented on a large scale, particularly in the financial services sector, where early deployments are delivering measurable gains in efficiency and productivity. For customers, this means faster service, more personal interactions, quicker resolution of everyday banking needs and less time spent navigating routine processes.
Take the customer acquisition phase as an example. Incorporating a small or medium-sized business traditionally takes many hours over several days, with relationship managers collecting documents, verifying business registration, carrying out Know Your Customer (KYC) checks, assessing credit risk and configuring products across various systems. That level of manual work increases the cost of acquiring each customer, which in turn limits how widely banks can serve the market, often forcing them to prioritize larger, higher-value customers over smaller businesses.
Now imagine compressing that process into about 20 minutes through automation using tools like biometric identity verification, automated Companies and Intellectual Property Commission (CIPC) lookups, real-time credit bureau integration and instant KYC orchestration. Suddenly the economy changes and banks can serve sectors that were previously not viable on a large scale, expanding their addressable market and opening the door to broader financial inclusion. For the SME owner, this means being able to open a business banking account in minutes rather than days, with much less time spent navigating paperwork and administrative processes.
Many banks in South Africa have already begun to invest in these technologies, particularly in business-critical operations, and their application in other areas of the banking value chain is expected to expand significantly by 2026.
For example, in September last year, Absa partnered with cloud-based customer relationship management platform Salesforce to bring its enterprise agent AI solution, AgentForce, to Africa for the first time, a development that has significant implications for the banking industry.
As part of this, three autonomous AI agents are being tested within the bank.
The first is a co-pilot designed to support relationship managers. Historically, most of their time was spent preparing for meetings rather than connecting with customers, with more than 75% of the workday spent reviewing customer data across various systems, analyzing recent transactions, identifying potential opportunities, and compiling engagement notes. Only about 25% of their time was actually spent interacting with customers. AI agents now automate that preparation by preparing pre-meeting briefs, pulling together customer information from multiple systems, uncovering relevant transaction patterns, and capturing notes after the engagement. This has effectively changed that ratio, reducing administrative time by between 75% and 90%, helping relationship managers focus more of their time on customers.
Another agent focuses on customer inquiries. It operates as a multilingual support assistant capable of answering common banking questions in 11 languages including isiZulu, Sesotho and isiXhosa. Early results show that approximately 40% of customer service queries are now resolved without human intervention, meaning four out of ten interactions can be handled immediately while maintaining response quality. In many cases, responses that used to take customers up to 30 minutes to receive are now delivered instantly, with a success rate of approximately 99%. The third agent focuses on monitoring and resolving technical issues in the system. Operating around the clock, it can automatically detect and resolve many issues, reducing the time taken to restore services and allowing teams to focus less on operational troubleshooting and more on supporting customers. Since its launch, the agent has already handled over 6,400 internal support queries with a success rate of approximately 96%, meaning the majority of issues have been resolved without the need to reach out to technical support teams.
These examples offer a glimpse of how agentic AI is starting to change the way banks operate and, more importantly, what customers can expect in the near future.
It is likely that this technology will be used more to expand digital onboarding, allowing customers to open accounts or access services more quickly and with less paperwork. Banks can begin to use AI to assess creditworthiness in new ways, analyzing transaction patterns and cash flow behavior rather than relying solely on traditional balance sheet information. For small businesses in particular, this may mean quicker access to financing and more active offers when their financial activity shows they are able to support additional debt. At the same time, many routine banking interactions are likely to shift to digital self-service channels, allowing customers to resolve simple requests quickly, while still maintaining access to human assistance when actually needed.
Agent AI is still in its early stages, but its direction of travel is already becoming clear.
As these systems mature and banks become more comfortable deploying them in everyday operations, customers are likely to experience banking that is faster, more responsive, and increasingly personalized. Much of this change may be happening in the background, but over time it will change how people interact with their banks.
Article written by Lindelani Ramukumba, Absa Business Banking Chief Information Officer and Interim Chief Digital Officer