Foundational AI models are trained primarily on Internet data — and Africa’s voices, languages, and cultural contexts are significantly underrepresented in that data.
This is the word from Linda Saunders, country manager and senior director: solution engineering Africa at Salesforce, grounded a recent roundtable discussion in a reality specific to the continent.
The real opportunity, she argues, lies not in the foundation model but in the context layer: the enterprise ability to localise AI using an organisation’s own data, conversations, and values.
Conversational platforms like Slack, notes Saunders, are treasure troves of organisational intelligence that most businesses have barely begun to mine.
“There is no putting this back in the box,” she warns. “The decisions we make today – the education we provide, the narratives we choose – will have far-reaching consequences. I would hate for us to look back and realise we chose the wrong narrative at the moment it mattered most.”
Ursula Fear, senior talent program manager at Salesforce, points to the skills crisis lurking beneath the AI opportunity.
AI curricula are evolving on four-month cycles; South Africa’s formal qualification frameworks operate on five-year timelines. That gap, she points out, is not a future problem – it is a present one.
According to Fear, learning must move into the flow of work itself, with professionals committing five to 10 hours a week to staying current.
Micro-credentials and free platforms like Salesforce’s Trailhead are democratising access to skills that were previously out of reach.
“The sweet spot is genuine human-agent collaboration – AI handling the repetitive, humans freed for what they do best. But that does not arrive automatically,” Fear says. “You have to build toward it deliberately, through continuous learning embedded in the work itself.”
She also called for a fundamental redesign of job architectures – away from static roles and fixed descriptions, toward flexible, AI-integrated workflows that reward adaptability.
Young AI natives entering the workforce, she notes, are a significant organisational asset: the challenge is building environments that can genuinely absorb and activate them.
Heldi Levy, change lead for the Salesforce programme at Standard Bank Group, highlights the fact that leadership commitment cannot be delegated – executives who do not visibly use AI tools send a clear signal that transformation is optional.
Alongside visible leadership, she identifies change agents – passionate early adopters embedded across every function – as the human connective tissue that makes adoption sustainable.
The key to unlocking that adoption is answering one honest question for every employee, Levy says. “Adoption is driven by one question: what is in it for me?
“When employees experience, directly, that AI removes the parts of their job they like least and gives them back time for meaningful work, resistance dissolves. You stop pushing and the organisation starts pulling,” she said.
Levy adds that, as AI embeds itself in operational decision-making, board-level ownership – not mere oversight – is non-negotiable. Individual responsibility matters equally; employees must understand both the power and the risks of the tools they use.