Serving Tomorrow’s Hybrid Customers

Irwin Lim
Deloitte Digital Leader
Deloitte Southeast Asia

Robert Gallagher
Technology, Media & Telecommunications Leader
Deloitte Thailand 

Tomorrow’s digital experiences must serve both humans and the AI agents that represent them. 

A profound shift in customer behaviour is emerging, as AI advances make it possible for customers to delegate routine decisions to personal AI agents that can browse, compare options, and take actions on their behalf. Research by MARKETECH APAC reported that 88% of consumers in Southeast Asia rely on AI for purchasing decisions . At the same time, many organizations are deploying their own AI agents to handle customer interactions, diagnose problems, and execute service workflows. However, a human touch remains essential. A meaningful subset of customers still demand empathy and reassurance from humans in moments of high stakes. Therein lies the premise of our point of view: tomorrow’s customers will be hybrid—sometimes human, sometimes their AI agent, and often both in a single journey. Hence, businesses must be hybrid too. We call this hybrid-to-hybrid CX.

This dynamic is visible across telecommunications, financial services, and retail — industries where digital and human interactions define customer relationships. The implications are profound. Businesses that understand this shift early — and design experiences for both humans and AI agents — will be better positioned to win customer trust, loyalty, and growth.

Also, the economic momentum behind this shift is significant. The global market for this new category of AI — known as "agentic AI," is forecast to grow from US$7.8 billion in 2025 to US$52.6 billion by 2030, at a compound annual growth rate of roughly 46%. Consider a few emerging examples:

In financial services, a customer may add stock to their watchlist. Then, an AI agent of their bank’s relationship manager monitors market signals and recommends when to make a sales call. 

In telecommunications, a customer’s personal AI agent can detect irregular connectivity. Without the customer lifting a finger, the AI agent flags the issue to the operator’s service agent, which orchestrates diagnostic tools, triggers resolution workflows, and initiates automatic compensation if service thresholds are breached. From the customer's perspective, the problem was fixed before they even knew it existed.
 

The design gap most organisations have not yet recognised

In the past, customer experience was built around two fundamental assumptions: the customer is human, and the interaction happens on a screen. Organisations invested heavily in glossy, human friendly interfaces. Yet, this is no longer sufficient. Hybrid-to-hybrid CX needs systems that are machine-readable, policy-clear, consented, and observable. This creates a significant gap for many organisations. But it is also a significant opportunity. 
 

Segmenting hybrid-to-hybrid CX 

In a hybrid, agent-mediated world, the question is no longer “What will a user click?” but “What can an agent reliably finish, and what should remain human-led?”. Hybrid-to-hybrid CX should adapt to customer intent and preference – specifically, whether they are assurance-seeking (human touch), autonomy-seeking (self-serve) or delegators (agent-first). The same customer can shift between these depending on the moment.

In hybrid-to-hybrid CX, there are four interaction archetypes, each differing in who leads the work and where humans intervene. 

1. Human-to-Human (human executes, agent validates)
   Proper in high-emotion, high regret cost, or trust-based moments where customer needs reassurance, explanation, and ownership.
2. Augmented human (human decides; agent advises)
   Customers interact with humans, while AI agents assist in the background by providing information and recommendations for faster handling with fewer errors. 
3. Human-to-agent (human orchestrates; agent operates)
   Suitable for repeatable intents where customers want speed and are comfortable delegating, but companies must control identity, consent, and risk.
4. Agent-to-Agent (human supervisors, agent executes quasi-autonomously)
   Suitable for high-volume, repeatable intents spanning multiple domains where fulfilment can be executed through APIs/events, while humans monitor guardrails and escalation when necessary. 
    

Four imperatives for building hybrid-to-hybrid CX

1. Pivot from data capture to data activation
   Building an always-on signal layer— one that surfaces relevant information instantly to both your AI systems and customers' delegating agents — is the foundation of effective hybrid CX. With this in place, outcomes include more relevant interactions and upsells, proactive remediation, and increased customer satisfaction. 
2. Make products and services agent-executable
   In the future if your products and services cannot be discovered and transacted by a customer’s AI agent, they are nothing. Organisations need stable, deterministic APIs for high-value intents and must express business rules as policy-as-code, with clearly defined inputs, side effects, and outcomes. 
3. Design experiences by target segment
   Not every customer interaction is equal, nor should customers experience the same thing. Organisations should design agent journeys to be configurable by customer preference and context. This improves retention by protecting high-touch moments and compressing handling time. 
4. Enable data sharing with consent
   Organisations should provide granular identity, consent, and delegation, plus a human-in-the-loop for oversight, explainability, and key approvals with clear governance.

Despite all its power, we believe the future of AI will remain human-led — with agents delivering speed and scale, and humans anchoring trust and relationship.