Digital first, not digital only: hybrid human + AI service

Service with AI alone still doesn’t hold up. Gartner predicts that, by 2027, half of the companies that planned to heavily reduce human service will give up on that plan, and 95% of service leaders intend to keep human agents. The path is the hybrid model: digital first, but not digital only.

What Gartner states

The forecast is in the June 10, 2025 statement, based on a March 2025 survey of 163 service and support leaders. The company cites the difficulty of hitting team targets with no human agent at all. Kathy Ross, senior analyst at Gartner, says in the statement that the human touch remains irreplaceable in many interactions and that the model where AI and agents work together is the most effective.

The statement recommends defining AI’s role, prioritizing strategic objectives, determining the human agents’ functions and using AI as a complement. It doesn’t bring numbers on cost or service time. The full analysis is paid and we didn’t consult it.

The Klarna case gives the forecast substance. The company said its assistant did the work of 700 agents and then went back to prioritizing human support. We cover it in the Klarna lesson.

Who handles what

Gartner doesn’t define a split by case type. The one we propose below is ours, and should be tested in your operation:

Case type	Who starts	Examples
Simple and repetitive	AI resolves alone	Second copy, order status, hours and address
Intermediate	AI starts and hands to a person when needed	Exchange, plan question, billing adjustment
Complex or high-value	Person from the start	Serious complaint, dispute, strategic customer

The ruler for each company changes with the product, the ticket and the customer profile. Start with few cases in AI’s column and expand as the customer score and recontact show it’s working.

Handoff without the customer repeating everything

The point where hybrid usually fails is the handoff from AI to a person. If the agent arrives without seeing the conversation, the customer tells everything again, and AI loses what it gained. The handoff needs to carry the history, a summary of the problem and what’s already been tried. That depends on a single customer record across channels, the topic of omnichannel service with AI.

The agent’s new role

In our reading, the agent now receives the cases AI didn’t resolve, which are the hardest. The work changes: fewer repeated questions, more complex problems and conversations with an irritated customer. That calls for training and a different metric, centered on resolution and score, not on the number of tickets.

How to start

1. List the 10 most common contact reasons and classify each in the table above.
2. Choose a channel for the test, preferably the one your customers use most. If it’s WhatsApp, the 10 precautions before automating help.
3. Define the handoff rules to a person and what goes along.
4. Measure score, recontact and resolution. The step by step is in how to measure an AI service agent.

To build this design with us, see sales and service with AI.

Frequently asked

Can you have 100% AI service?

Gartner’s forecast (Jun/2025) is that half of the companies that planned to heavily reduce human service will abandon that plan by 2027. Gartner itself recommends AI as a complement, not a substitute.

What is hybrid service?

It’s service in which AI and people split the cases. AI resolves the simple and the repetitive, and a person steps in when the case is complex or the customer asks.

How many service leaders will keep humans?

In Gartner’s survey of 163 leaders (March 2025), 95% intend to keep human agents to define AI’s strategic role.

How do you keep the customer from repeating everything when talking to a person?

With a single customer history and an automatic summary of the conversation for the agent, sent at the moment of handoff.

Sources

• Gartner. 50% of Organizations Will Abandon Plans to Reduce Customer Service Workforce Due to AI, 10/06/2025. gartner.com
• Tech.co. Klarna reverses AI overhaul, 20/05/2025. tech.co