Somewhere in the next few years, a meaningful slice of your customers is going to stop being human. Not in any spooky sense, the principal behind the purchase is still a person with a wallet and a want, but the actual buyer, the entity comparing your product against competitors and clicking "add to cart," is going to be an AI agent acting on that person's instructions.
This is already happening at the edges. Procurement teams are piloting agents that source components. Consumers are asking ChatGPT which medical aid to take out and which router to buy. "Agentic commerce" stopped being a conference buzzword and started showing up on Shopify's product roadmap.
If you run marketing for a brand, whether that's a challenger FMCG punting through Checkers, a fintech trying to win SMEs off the big four, or a B2B SaaS selling into the JSE top 40, you need to think carefully about what changes when your buyer has no amygdala.
Most of your advertising spend is aimed at a system the AI doesn't have
Modern marketing is, to a surprising degree, psychology dressed up in graphic design. Emotional storytelling, aspirational imagery, packaging that triggers a dopamine flicker at the shelf, scarcity cues ("Only 3 left!"), social proof, celebrity endorsement, colour theory, almost the entire toolkit is optimised for a buyer with a limbic system, cognitive biases, and limited attention.
An AI agent has none of that. It will not be seduced by a beautifully shot advert of a bakkie driving through the Karoo. It will not feel FOMO. It will not give your brand a pass because it saw the Castle ad during the Springbok test. It reads specifications, compares them against a brief, and makes a decision.
This does not mean emotional marketing dies. It means it retreats to the categories where a human is still making the call, identity goods (what a person wears, drives, displays), gifts, experiential purchases (restaurants, holidays, concerts), and high-trust personal decisions people refuse to delegate. Expect above-the-line spend to concentrate sharply in these areas, and expect the budgets for commoditised purchases, insurance, utilities, consumables, office supplies, most B2B procurement, to redirect somewhere else entirely.
That somewhere else is a discipline we don't quite have a name for yet
Call it agent optimisation, machine-readable marketing, or B2A (business-to-agent), the point is the same. When an AI is the buyer, your job is to be the product it chooses.
In practice this means three shifts.
Structured data becomes your shop window. The product feed, the API, the specification sheet, the JSON-LD markup on your category pages, these are now customer-facing assets with the weight that TV creative used to carry. If your product details are locked inside a PDF brochure, or worse, inside the heads of your sales team, AI agents can't see you. You're invisible in exactly the way that not having a website made you invisible in 2005.
Verifiable claims beat clever copy. An agent weighing two fridges doesn't care that yours is "energy-smart", it wants the kWh/year figure, ideally attested by a credible third party. Marketing copy full of adjectives becomes noise. Marketing copy backed by data the agent can cross-reference becomes signal. South African marketers have a slight head start here because we've had to deal with Consumer Protection Act scrutiny and ASA rulings for years; the discipline of substantiating claims is already part of the job. It's about to become everything.
Presence in the training data matters. If an agent is asked "what's the best short-term insurer in South Africa for a young driver," it is drawing on what it has read, forums, review sites, news articles, comparison platforms, the structured web. Brands that invested in earned media, Wikipedia presence, authentic community discussion, and relationships with comparison platforms will keep showing up in that answer. Brands that relied purely on paid performance marketing will not. This is a slow-moving, compounding advantage, which is exactly the kind incumbents usually struggle to build quickly.
The death of inertia is a massacre of loyalty programmes
A great deal of South African marketing, and South African retail margin, depends on human laziness. You stay with FNB because switching is admin. You renew your short-term cover because the alternative is a morning on the phone. You keep buying the same laundry powder at Pick n Pay because comparing unit prices is not how you want to spend your Saturday.
AI agents have no laziness. Every renewal is a fresh tender. Every repeat purchase is a re-competition. The implications are brutal: customer acquisition costs rise, lifetime value falls, and the whole arithmetic of loyalty spend gets redone.
Loyalty programmes themselves, Smart Shopper, Clicks ClubCard, WRewards, Vitality, will not disappear, but they will need to justify themselves in transparent rand-and-cent terms. An agent can calculate, in milliseconds, whether your points are actually worth more than the competitor's straight discount. Programmes that rely on psychological overvaluation ("I've got 14,000 miles, I can't let them go to waste") will be stripped down to their real economic value. Programmes that deliver genuine, measurable savings will survive. The middle, which is most of them, will get eaten.
Dark patterns stop working, and it's about time
Drip pricing, decoy products, confusion marketing, basket add-on tricks, deceptive urgency, artificial scarcity, the whole catalogue of manipulative pricing that has grown like kikuyu across local e-commerce, becomes near-useless against an AI buyer. Agents read total cost. They ignore countdown timers. They don't buy the premium warranty because it was default-selected.
Marketers who've built careers around squeezing extra margin through interface dark patterns are going to have a hard time. Marketers who've built brands around genuine value are going to have a much easier one. The Competition Commission will, with any luck, have less to do.
The new gatekeepers are the agent platforms themselves
Here is the development that should make every marketing director uneasy: whoever controls the dominant buying agents controls enormous purchasing power. If most consumers eventually shop through one of three or four big AI assistants, those platforms become the new shelf space, the new Google, the new supermarket category buyer. They will decide, through their ranking logic, their training data, their commercial arrangements, which brands get surfaced.
We've seen this film before with search engines, app stores, and social platforms. Expect the sequel to include pay-for-placement schemes dressed up as "certified partner" programmes, opaque ranking algorithms that marketing teams spend fortunes trying to reverse-engineer, and eventually regulatory intervention that arrives several years too late. Smart marketers will start cultivating relationships with the agent ecosystem now, while the terms are still being set.
What to do on Monday morning
None of this means ripping up next year's plan. It means tilting it. A few practical moves.
Audit your digital estate from a machine's point of view. Can an AI agent actually find, parse, and verify your product information without human help? If not, that's where the next round of web budget goes.
Start thinking about your brand as a specification, not just a feeling. "What does our product objectively do better, and can we prove it?" needs an answer that fits in a structured table, not only on a mood board.
Reduce your dependence on channels that only reach humans. Paid social, TV, out-of-home, and print still matter, but treat them as tools for the human-decision portion of your category, not the whole of it.
Get serious about earned presence, reviews, expert coverage, comparison platforms, credible third-party validation. This is your organic distribution in the AI era.
And watch the agent platforms the way marketers a decade ago should have watched Facebook's algorithm changes, because the next big shift in where your customers come from is already being written, and the people writing it are not in your industry.
