Event Archive

Linda Saunders, Country Manager at Salesforce South Africa

Every few years, the technology sector is swept up by talk of “creative destruction”. A new innovation arrives, predicated on the absolute obsolescence of the status quo. We were told the iPhone would finish off the PC; that cloud computing would mean the end of the datacentre; and that blockchain would render traditional banks irrelevant.

None of them came to pass. To paraphrase Mark Twain, reports of enterprise software’s death have been greatly exaggerated.

AI might be able to write code, but it is a massive leap from writing code to running a business on it. Many leaders mistakenly believe AI models can produce fully functional, enterprise-grade software ready to deploy. They cannot.

AI is excellent at churning out small tools for specific tasks, like summarising meetings or drafting emails, but that is vastly different from designing an enterprise-wide system that remains stable, secure, and cost-effective. If you ignore that distinction and stack automated tools haphazardly, you end up with a “Frankenstein” jumble of systems that are harder and more expensive to run than a single, well-designed platform. True competitive advantage comes from a unified architecture that bridges four key pillars:

● Systems of Context: Your unique data and history

● Systems of Work: Your business logic and processes

● Systems of Agency: Your AI agents that reason and act

● Systems of Engagement: The interface where your users and customers interact

The gap between a powerful language model and actual business outcomes is almost always a data problem. If the data feeding a model is inconsistent or ungoverned, the output is worthless. However, the “missing link” isn’t just having data; it’s having shared metadata.

Unlike closed ecosystems that require you to move all your information, a collaborative platform allows for “zero-copy” integration, maximising investments you’ve already made in external data lakes or proprietary LLMs. By using metadata as a common language, Salesforce ensures AI respects existing security permissions and business rules. This technical “spine” creates the accuracy and speed necessary to turn AI from a boardroom demonstration into an operational reality.

With a cloud platform, economies of scale mean many customers share the cost of the heavy lifting. The vendor builds and runs the core system once; everyone benefits from the same upgrades, resilience, and security.

If you try to “build it yourself”, costs add up quickly. You pay for every bit of AI processing, you’re wiring together separate tools, and when something goes wrong, even a minor glitch turns into a full-blown firefight for your tech team.

Anxiety about automation and job losses is legitimate. At Salesforce, however, our experience suggests growth in high-value work. Since deploying Agentforce, our enterprise agentic AI solution, we’ve seen hundreds of employees redeployed to high-growth areas where human judgement and expertise create the most value.

AI is also creating entirely new roles. We are now hiring for positions that barely existed a few months ago: deployment strategists, AI conversation designers, and AI architects. Despite deep AI integration, our global team has increased by nearly 10 000 people since 2022.

We see the impact of this growth, particularly in South Africa, where businesses are using automation to bypass traditional infrastructure bottlenecks and leapfrog into more sophisticated, data-led operations. AI isn’t wiping out roles; it is shifting the burden of labour. By delegating repetitive tasks to digital agents, we allow people to focus on complex problem-solving and strategic growth. Instead of asking whether automation will replace enterprise software, we should ask what it will turn enterprise software into. We’re moving towards “service-as- a-software”, systems that can make decisions, trigger actions, and coordinate work across teams using your data and rules.

At that point, growth will be limited less by how many people you can hire and more by how much trusted computing power you can put behind your most important processes. Automation isn’t the replacement for enterprise platforms; it’s the reason they matter.

We can all get behind a future where a morning’s work produces a full day’s output. That’s not about working less; it’s about digital agents taking on the grind so people can focus on creativity, strategy, and success.