Date:

Generative AI raises the governance bar: Viewwave wins at the SA AI Awards

As banks move from pilots to generative and agentic AI, the winner of this year’s banking and finance award argues that evidence, not enthusiasm, should decide what reaches customers.

Viewwave has won Best AI in Banking and Finance at the inaugural South African AI Awards, held at The Galleria in Sandton on 13 November 2026. The company was recognised for making responsible AI governance work in practice inside South African financial institutions. The category judged entries on criteria including risk reduction, explainability and governance controls, data security and regulatory alignment.

Their argument is simple: a model that performs as intended is not necessarily one an institution can defend. The distinction becomes sharper with generative AI, where outputs vary, failure modes are harder to predict and systems increasingly act with some autonomy.

“With generative AI you can’t test every answer in advance,” said Dr Johan du Pisanie, Principal at Viewwave. “What you can do is define what the system is allowed to do, test how it behaves at the edges, put controls around it and keep the evidence. That’s what lets a risk committee say yes with its eyes open.”

From pilots to production

In 2026 Viewwave worked on a CRO-sponsored engagement at a major South African financial institution that extended its AI governance architecture to generative AI, strengthened its acceptable use standard and built AI validation and assurance into its operating model. Management received a prioritised view of gaps and a sequenced 12-month improvement roadmap.

At a global bank, it built a reusable testing, evaluation, verification and validation (TEVV) capability for generative AI, with a risk taxonomy, AWS Bedrock-enabled controls, monitoring and readiness evidence. At a Fortune Global 500 financial institution, it tested real-world and edge-case behaviour and quantified failure modes before the system went into production.

Six questions before go-live

Across these engagements Viewwave uses six questions to decide whether an AI system is ready: what exactly is being governed, what can go wrong, which controls are required, how the system was tested, who owns the residual risk, and what evidence supports approval.

Viewwave works with its partner, RiskAI, to bring these solutions to clients.

“An AI agent that can act, not just answer, needs what any new employee needs: clear limits, supervision and someone accountable for its decisions,” said Rohit Kumar, Founder and CEO of RiskAI.

A national gap

The awards’ assessment of entries found that scores in the AI ethics and governance category ranged from 20 to 84. With the national AI policy being redrafted and expected to give sectors more responsibility for their own rules, institutions that build governance now will be better placed for whatever follows.

“Responsible AI isn’t a separate project,” said du Pisanie. “It’s how you get AI into production without surprises, for the institution and for its customers.”

Share post:

spot_img

Popular

spot_img

More like this
Related

The twin fault lines: how broken escalators and spatial traps are redefining unrest in Europe and South Africa

Across Western Europe and South Africa, many people are...

Curate introduces three new global equity feeder funds

Funds provide advisers and investors exclusive access to world-leading...

Black Friday 2026 will test retailer payment platforms and resilience

This Black Friday-Cyber Monday South African retailers face a...

Business needs both logic and magic. The trick is knowing when to switch off

As we enter the final quarter of 2026, I...