Customer Service Agents vs Outsourcing
Who answers when you dial for help tomorrow?
Interactive voice response. Overseas call centers with American names but non-American English. Long wait times. Something has gone wrong if we need to engage with a company's customer service. Companies don’t want us to engage with customer service either. They want to prevent, or deflect, customer service, a cost center.
I do my best to smash #0, #9, repeat “human” ad nauseam, or anything else I can think of to skip to a human. Soon, my next customer service inquiry will likely feature an AI agent of some type. And, I might even be happy about it!
AI-Driven Agent Platforms in Customer Service
AI agents are carving out a wedge in customer service. Businesses are open to this because they are already spending billions of dollars outsourcing customer service to Business Process Outsourcing (BPO) providers specializing in customer experience. This can be call centers, online chat, or email assistance.
AI intelligence layers, such as those offered by Microsoft, Salesforce, and ServiceNow, are advancing into agentic platforms capable of handling lower-level customer service inquiries. These platforms mark a shift from traditional "systems of record" to "systems of action" or "systems of work."
ServiceNow’s diagram positioning a system of work, before AI agents.
Unlike older chatbots or IVR systems, modern agentic platforms are easier to implement. They can use existing documentation, often stored online or within an organization, to understand and manage workflows. By employing retrieval-augmented generation (RAG) techniques, these systems can streamline processes without requiring extensive manual pre-configuration.
For example, instead of pre-defining responses to every scenario, these platforms learn from workflows and documentation, allowing them to act on tasks more autonomously. They integrate into systems like Salesforce to access stored user flows and improve performance. These advancements enable better handling of customer queries and significantly enhance the user experience.
Economic Impact on BPOs and Customer Experience Providers
Enterprises and customer service-oriented BPOs are piloting these agentic customer service platforms. The economics of these platforms align closely with current BPO cost structures. For instance, Salesforce's quoted price of $2 per interaction corresponds to the effective cost of handling a five-minute customer service call through BPOs.
Overseas customer service might cost $15 an hour.
The average interaction takes about 5 minutes.
That is ~$1.25 in labor cost to the BPO. BPOs typically operate with gross margins between 30% and 50%.
Putting a customer cost at $2 to $2.40, right inline with Salesforce pricing!
By adopting AI systems, they can optimize costs and potentially improve these margins. For example, AI can help BPOs manage high turnover by reducing the need for immediate replacements when employees leave. Over time, this gradual transition enables cost savings without large-scale layoffs or workflow disruptions.
Many BPOs are already investing in generative AI solutions to maintain competitiveness. Companies like TaskUs and Teleperformance have developed in-house AI tools, such as AssistAI (formerly known as TaskGPT), to enhance their offerings. These investments have positioned them as early adopters in the shift toward AI-driven workflows.
Adoption Patterns and Challenges
To understand how automation, AI, and agentic processes are impacting the workforce, I did numerous customer calls. I found, despite their potential, AI solutions are not typically implemented as full-scale replacements for existing systems. Companies tend to adopt them incrementally, prioritizing cost savings while avoiding abrupt workforce changes.
This approach often limits the achievable improvements to about 30%, compared to the 80% efficiency gains possible with a complete system overhaul.
Adoption varies by company maturity, the complexity of customer interactions, and the nature of the product or service. Most AI implementations begin with handling tier-one support tasks, such as answering common questions or resolving basic issues. Human agents are then freed to focus on more complex workflows.
Revenue-Driven Use Cases and Retention Benefits
AI in customer service is not limited to cost savings. Some implementations directly contribute to revenue by improving customer retention or upselling opportunities. For example:
Retention: AI systems can help resolve cancellation requests more effectively, reducing churn.
Upselling: Agents equipped with AI tools can identify opportunities to cross-sell or upsell, increasing average order value (AOV) and customer lifetime value (CLV).
Revenue-driving applications are often adopted more quickly than cost-saving solutions because they offer measurable top-line benefits.
Future Trends and Competitive Pressures
AI agent platforms are developing rapidly. Salesforce's AgentForce, for instance, grew from 200 clients1 to 1,000 within a month2, reflecting the speed of adoption. However, these platforms are not solely about technology. Successful adoption also depends on building trust and demonstrating a clear return on investment (ROI).
Enterprises often hesitate to adopt new AI solutions from startups, preferring to wait for offerings from established vendors like Salesforce or ServiceNow. This creates a competitive challenge for smaller providers.
Another consideration is the positioning of AI systems within the customer service ecosystem. Solutions focused on deflecting calls—handling queries like “Where is my order?” or “How do I log in?”—target cost centers. In contrast, those that enhance customer retention or upselling potential address revenue-generating opportunities.
Conclusion
Companies will adopt a blended approach to customer service, trying to be as parsimonious as possible. The customer service stack might look like:
Self-service tools for deflection
AI or IVR routing (lower cost than agents)
AI agents
Outsourced call centers
Full time customer service employees
The increased deflection, and efficiency, thanks to AI, will reduce human staffing in these roles. The changes won’t come all at once, but will roll out over time. People (employees and customers) do not want to switch from humans to AI all at once. Gradual change will make it easier to accept.
However, the change will come. The impact will be real. Companies will save money and we will all get service faster. Customer service is a big vertical to target, but these changes will come to other out sourced areas as well (IT, Finance, HR, etc). Traditional BPO incumbents will try to adapt to the change, but will likely see shrinking markets. Markets that companies like Salesforce and ServiceNow will only be too happy to shrink.
Salesforce FY Q3 2025 Earnings call held on December 3rd, 2024 - https://seekingalpha.com/article/4741913-salesforce-inc-crm-q3-2025-earnings-call-transcript
Agentforce 2.0 talk https://www.reuters.com/technology/artificial-intelligence/salesforce-closes-1000-paid-agentforce-deals-looks-robot-future-2024-12-17/