Personal agents are starting to file complaints, dispute bills and call contact centers on their owners' behalf—adding volume and complexity that service leaders need to address now.
Personal agents are already disrupting service
This is just the beginning
Customers’ use of AI has evolved from fabricated product images supporting return claims to carefully scripted calls demanding refunds or invoking lemon laws. Now, service leaders face a potential tsunami of personal AI agents that customers can deploy with little more than a simple instruction.
These agents are already calling, texting and chatting on customers’ behalf. The early impact is a glimpse of what could follow as adoption grows. Contact centers have spent the past several years deploying AI internally for routing, agent assistance and self-service. This next wave is arriving from the outside, directed by customers—and service leaders need to respond now.
Consider a refund that would have typically taken three phone calls. The order arrived damaged, the retailer's app offers store credit only, the written policy allows a cash refund if you ask within 14 days, and the chat widget occasionally crashes before connecting to an available representative. A personal AI agent doesn't get impatient. It reads the policy, keeps retrying the chat until it gets through, quotes the clause back to the rep, and asks for the refund—repeatedly if necessary—until it achieves its goal.
Customers have a new, increasingly capable and persistent representative
Meta launched Muse in September as an agent that "can open a browser, fill out forms, and negotiate on their behalf." "Lowering a bill" was one of its launch examples, and by October 21, 2026, Meta announced Muse had over 10M users in the US and Canada.¹ In the same month, a 14-person startup called Instinct raised $1 billion at a $10 billion valuation for an agent that books flights and hotels, calls restaurants that don't take online reservations and, in its own words, sorts out problems with your cable bill. These are early numbers from companies with every reason to round up, but the direction isn't in doubt.
Service will require new ways to measure value as personal AI volumes grow
In 2025, the National Customer Rage Survey found that 77% of U.S. households had a problem with a product or service in the past year, but 68% said complaining took real effort.²
If a customer expects to spend an hour on hold when they reach out to an organization for support, they might decide not to pursue a disputed fee, a confusing line item or a lapsed warranty after all. A personal AI agent likely won’t make that trade-off. It queues the task and sees it through to an acceptable conclusion. Once customers can delegate these tasks to AI agents at scale, the incoming volume to your service center is likely to climb. However unintentional the existing friction for customer service interactions is now, it does drive down contact volumes, and organizations need to prepare for a new, higher-volume reality.
Today's service leaders must plan for the new volume of demand types that AI assistants can introduce. These could include duplicate submissions, more persistent contact methods or even potential fraud from bad actors.
Traditional service metrics do not clearly reflect the new trends that personal AI agents can introduce as they build momentum in the market.
Contact center metrics were not built for a caller that isn't the customer. Updated success measurements will each need to accurately recognize and separate AI agent-originated contact instances from those of human customers.
New metrics will also be required in the new service regime. As AI assistants increasingly make purchasing decisions on behalf of consumers, service channels that work well with these new AI audiences will become a key brand asset. An AI agent can remember which organizations provide reliable pathways to resolution and can steer its human owner accordingly in subsequent conversations. This means that whether an AI agent finds your contact center easy to navigate will become another key performance indicator (KPI) for your service organization.
Turning priorities into action
Update your operating model
Many organizations split the responsibilities underpinning service between various business and technology stakeholders across the enterprise. Service organizations that prosper in the personal AI era will start by making four shifts in their operating model:
Establish shared governance and decision ownership between the business, IT, legal and cyber teams so the service perspective shapes the organization’s strategic AI decisions rather than inheriting them.
Appoint one leader accountable for AI-facing channels; hire people with current AI skills and empower them to drive the technical changes required.
Revisit capacity plans to account for induced volume and queue congestion from callers that don't abandon.
Refocus success metrics on verified resolution and cost per resolved task, measured separately for human and agent-originated contacts.
Set new business rules for AI arriving in your traditional channels
Service leaders will also need to face a new question: Should we prioritize AI agents in our queues last?
The instinct to protect human access is sound, and so is the counterweight, which is that an authorized AI agent represents a real human customer with a real problem: a queue that penalizes it penalizes them. Define priorities and service commitments for agent-originated contacts explicitly rather than letting the routing engine decide by accident.
You will also need to enable the frontline to handle personal AI agents. Give employees clear procedures for handling AI agents instead of leaving them to improvise on the call.
Lastly, remember that AI personal assistants may eventually become their own customer base. Providing good AI service could inform and incentivize future AI purchases if a pattern of efficient service becomes associated with your brand.
Accelerate investment in core AI agent infrastructure
The foundational layers needed to drive an effective agentic experience for customers and employees include Data, Enterpries APIs, and Specialized Agents.
Data
A strong foundation lies in both the transactional data for a given customer and the policy and procedure details that govern how to execute a transaction. Together, they establish the baseline for an effective agentic service orchestration engine.
Enterprise APIs
Once customer and policy data is organized, expose it to internal AI agents through a consistent set of APIs. Keep those APIs and systems private; let the customer-facing agent handle requests and call the APIs only after verifying the customer’s identity and the personal agent’s authority. For example, the order management system should remain private, while an authorized agent could use it to process a return or exchange.
Specialized agents
The next layer up is your library of internal specialized AI agents, each with access to the procedure data and APIs relevant to its role. These agents can resolve a range of service requests autonomously across channels.
Build additional layers that only AI-to-AI service needs
Building your new service architecture will require two additional layers on top of the core service infrastructure. The three core AI service layers above support any copilot, chatbot or headless channel. Two additional AI-to-AI layers enable personal AI agents to efficiently and safely carry out actions on behalf of a customer.
Looking ahead
Service spent a decade teaching its AI to talk to customers, but the tables are starting to turn. Tomorrow's successful service organizations will increasingly blend digital and human ecosystems: agents serving agents, humans serving humans with AI support, and hybrid experiences that cross actors, channels and trust boundaries without friction.
Service organizations that do not rapidly adapt may find themselves overwhelmed by new service demand. Those that do may unlock new revenue by serving agentic customers that value good service even more than the humans they represent.
Next in this series
Upcoming topics in Deloitte Digital’s Future of Service trends include:
1. Cyber and safe AI. How do you authenticate personal AI agents and audit their behavior?
2. Headless agentic orchestration. How do you orchestrate customer and employee experiences across channels to provide seamless service?
3. AI economics and open models. What does a resolved request cost when a machine asks and a machine answers? Where can open models change the math?
4. Workforce. How do you prepare your frontline to respond to personal AI agents?
5. Trust. How do you win the trust of customer engagement with AI?
6. Field service and voice. What happens when the agent on the line is dispatching a technician, and/or is the technician's personal agent?
Authors
Mike Brinker
Principal, US Service Domain Leader
Deloitte Consulting LLP
Erik Golden
AI Transformation Architect
Deloitte Consulting LLP
D.C. Mungo
Senior Manager, Service Strategy & Transformation
Deloitte Consulting LLP
Marisa Lorch
Senior Consultant, Service Strategy & Transformation
Deloitte Consulting LLP
Transform customer service with an AI-first approach
Turn agentic AI potential into bottom-line performance
Reinventing customer experiences with solutions that inspire trust
SOURCES
