The JSE is testing AI agents on market data, but only for advisory and informational work. Alicia Greenwood of the JSE set out the approach in an article for TechCentral.
Greenwood said one project is testing how agents could securely question market data, starting with equity-trade data. Before these tools can grow, the exchange is testing the identity, access, data-governance and audit controls around them.
“The governance challenge changes fundamentally when AI moves from providing information and recommendations to taking actions on behalf of an organisation,” Greenwood wrote.
Greenwood said this matters for financial markets because they depend on confidence in their infrastructure, information and decision-making. A tool that helps a worker research or draft something carries a very different risk from an agent that can act inside a market process.
As AI gets more freedom to act, its limits must get stronger, Greenwood said. The same applies to the ability to audit its actions and to human accountability for the results.
Greenwood cited Boston Consulting Group (BCG), which found that 42% of companies expect AI agents to act without human approval by 2030. Only 5% have the relevant controls in place, according to BCG.
The JSE already uses AI in other parts of its work, Greenwood said. General productivity tools are available across the organisation. AI is also helping to modernise one of the exchange’s core systems by analysing complex old code, working out business requirements and creating test cases.
Greenwood said the challenge is turning these gains into lasting business results. A McKinsey survey for 2026 found that 80% of respondents believe AI has improved their own productivity. Only 37% reported any impact from AI on earnings before interest and tax.
Greenwood also pointed to figures shared at the JSE Future Competitiveness Forum. One large South African financial services group said about half of the servicing interactions in one of its businesses moved from voice to AI within three months. The group said AI also contributed to a 95% drop in client fraud losses in certain areas.
A technology investor at the forum cited about US$100-million in AI-related cost savings at one of its portfolio companies. Another investor said its AI costs rose 58-fold in a year as use grew. Later changes cut those costs by 90% without lowering product quality.
BCG estimated that corporate AI spending has more than tripled in two years to 3.3% of revenue. It said 80% of that spending now sits outside company IT budgets. Greenwood said AI should compete for capital on the same basis as other strategic investments.
Greenwood said the JSE starts with the business outcome before it chooses a technology. The exchange has deferred AI projects where questions about data, legal matters, costs or the business case were not resolved.
Source: this article is based on reporting by TechCentral. Image: TechCentral. Written with the help of AI and published by the Tzaneen Voice Technology desk. See our Editorial Standards.
Werner Jacobs covers technology for Tzaneen Voice. Werner reports on mobile networks, data prices, internet access, smartphones, apps, AI, cybersecurity and South African tech companies. Each story explains what a new product, price change or policy means for ordinary users in South Africa.