Customer operations after the AI wave: redesign around what AI does well and badly

Two things are true at once in customer service. Customers are using AI tools more freely than most companies realise. And customers are losing patience with companies whose own AI stands between them and a person who can solve their problem.

Leaders who design for both are likely to have a stronger service operation than those who treat AI purely as a headcount lever. The latest research gives a useful picture of the customer's side.

What customers say

In a press release dated 4 August 2026, Gartner reported a survey of 3,566 B2B and B2C customers, conducted in February and March 2026. Its findings are worth setting out plainly.

Three details deserve attention. First, half of customers find GenAI makes their interactions easier, so the technology is welcome when it works. Second, in their most recent service interaction, customers were about three times more likely to use a third-party tool such as ChatGPT, Gemini or Copilot than a chatbot provided by the company. Customers are already bringing their own AI. Third, when customers unwilling to use AI were asked what would change their minds, the most common answer was the ability to switch to a human agent.

What service leaders are planning

Gartner's December 2025 survey of 321 customer service and support leaders, conducted in September and October 2025, shows the other side. Ninety-one per cent were under executive pressure to implement AI, not only for efficiency but to improve customer satisfaction. More than 80% expected to reduce agent headcount within 18 months, mainly through attrition, hiring pauses or layoffs. Yet nearly 80% planned to move agents into new positions, 84% were adding new skills to agent profiles, and 58% planned to upskill agents as knowledge management specialists who review and curate AI-generated content.

A forecast frames the ambition. In March 2025, Gartner predicted that agentic AI would autonomously resolve 80% of common customer service issues without human intervention by 2029, with a 30% reduction in operational costs, as reported by CX Today. These are predictions, with no published method, and should be read as direction rather than plan.

Principles for redesigning the operation

1. Do not make AI the mandatory first step

Gartner's own recommendation is that GenAI should not be a compulsory first stage for every issue; forcing customers through repeated unsuccessful attempts makes them less likely to return to the tool. Use AI to gather information, understand intent and attempt a resolution only when confidence is high, with a clear path to a person.

2. Give the human the context

When a customer does reach an agent, the agent should already have the conversation, the customer's history and the AI's suggested next step. Making a customer repeat themselves is the fastest way to turn a neutral experience into a complaint.

3. Keep knowledge in excellent shape

AI answers are only as good as the knowledge behind them. Gartner's finding that many leaders plan to turn agents into knowledge specialists reflects this. Someone must own the articles, retire the outdated ones and correct what the AI gets wrong.

4. Automate transactions in the background

Gartner recommends conversational, action-oriented support with traditional automation completing the transactions behind the scenes. A customer asking to change a delivery address wants it changed, not a paragraph about how to change it.

5. Measure resolution, not deflection

A conversation that ends because the customer gave up is not a success. Measure first-contact resolution, repeat contacts, effort and satisfaction, alongside cost per contact. If cost falls while repeat contacts rise, the operation has simply moved its workload.

6. Treat the people side as seriously as the technology

If you reduce headcount, do it honestly and plan the new roles in advance: quality reviewers, knowledge owners, specialists for complex cases. The staff who know the customers best are the ones who can teach the AI where it is wrong.

What this means for organisations in Rwanda and across Africa

Our view, not Gartner's: the survey populations are global and we do not have equivalent African data, so treat this as a design brief and test it with your own customers. Many African markets are mobile-first, multilingual and conversational by habit. WhatsApp and voice channels are often primary, and local-language and code-switching support may matter more than a polished web chatbot. The principles still hold: be useful, resolve quickly and keep a person available.

Start small. Pick the three contact reasons that make up most of your volume, redesign them end to end, put the measures in front of the team each week and extend only what works. A customer operation built that way becomes more capable with each AI improvement, because the foundations (knowledge, process, measurement and people) are already in place.