I came across an interesting article this week, something that passed my ear.

In May 2026, Huawei and Konza Technopolis launched an AI powered contact centre at a digital hub in Bondo, Siaya County.

Twenty young people were being trained to run it, not to compete with the AI, but to work alongside it. The platform drafts responses to customer queries across chat, web, and phone. The trainee still decides what actually gets sent.

That is Kenya’s opening move in a debate global companies keep getting wrong at a much larger scale.

Job killer or customer experience enhancer. Siaya, part of the Jitume Digital Hubs programme that has now reached 290 hubs and over 400,000 young people, is built around a specific answer to that question: AI generates, a human decides.

It is a small pilot next to what Commonwealth Bank of Australia (CBA) and IKEA have already done with the same category of technology.

But both of those companies made the identical choice Siaya is now making, in opposite directions, and I think their outcomes are worth understanding before Kenya’s own BPO sector scales much further on this model.

Where the same bet went wrong

In 2025, CBA cut 45 customer service jobs and replaced them with an AI voice bot. Call volumes did not fall the way the bank expected.

They rose. Staff described management scrambling for overtime and pulling team leaders onto the phones to cover the gap.

CBA reversed the cuts within weeks and admitted it “did not adequately consider all relevant business considerations.” The Finance Sector Union was blunter: “CBA has been caught out trying to dress up job cuts as innovation.”

That should have been the end of it. It wasn’t.

In 2026, the same bank cut hundreds of contractor jobs at a call centre in Johannesburg, South Africa, run by an outsourcing firm called Nutun, saving an estimated tens of millions a year, while adding over 140 roles at its own centres in Australia.

The apology was domestic. The cut that followed was not.

That Johannesburg centre is not a hypothetical for us. It is the same kind of contract much of the region’s BPO sector competes for, work from banks and telcos abroad, priced on cost, vulnerable the moment an AI rollout needs a savings figure to justify itself to someone back at head office.

Where it went right

IKEA’s AI assistant, Billie, now handles 74% of customer inquiries, up from 47% a couple of years ago.

It answers store hours, return policies, the routine volume every contact centre drowns in.

IKEA did not treat that as a reason to shrink its team.

It retrained roughly 8,500 call centre employees, through a five to six week program, into resolution specialists for complex problems and design consultants who help customers plan kitchens and living rooms and sell them the products to do it.

Customer happiness scores went from 60% to 89%.

The remote sales channel those reskilled employees now run generates roughly €1.3 billion a year, growing 15 to 20% annually, IKEA’s fastest growing sales line.

Billie deals with pet policy. A human still decides whether your hexagonal countertop with a built-in stove is actually feasible.

What actually made the difference

Strip away the branding and CBA and IKEA made four decisions differently, and Siaya’s design is a direct bet on IKEA’s version of each one.

What they were solving for. CBA led with a savings figure, announced before the customer experience had been tested. IKEA asked what these employees could do that Billie could not, and built toward that instead.

What they let AI own.

IKEA drew a hard line at judgment. AI took the routine, low stakes volume.

Anything requiring taste, empathy, or a real decision stayed human.

CBA let AI take the whole first line, including the moments too complex or too emotional for a script, which is exactly where it broke. Siaya’s platform is built the same way IKEA’s is: the AI drafts, the trainee decides.

What happened to the people freed up. IKEA invested in a real training pathway with revenue accountability behind the new roles.

CBA simply removed the seats, so when the AI proved insufficient there was no one left to absorb the gap.

Whose jobs absorbed the risk when the plan did not work.

When CBA’s AI failed at home, the jobs came back. When the bank needed the savings anyway, it found them in Johannesburg, on contracts with less protection and less scrutiny. That is the default pattern whenever cost pressure meets an AI rollout that isn’t ready. The job treated as disposable is rarely the one closest to head office.

What this means here

So is AI in the contact centre a job killer or a customer experience enhancer? Both, depending entirely on what a company decides to protect when the savings pressure hits.

CBA protected a cost target and lost the trust of its own frontline before finding, thousands of kilometres away, a place cutting would attract less pushback.

IKEA protected 8,500 careers and built a billion euro sales channel doing it.

Siaya’s trainees are Kenya’s first answer to which of those two paths this sector chooses. It shows up in something as small as who decides what a customer actually receives, the AI or the person sitting behind it.