"Service as a Software" in the Customer Service Industry? Not really.
There’s been a lot of buzz in the VC ecosystem around the new "Service as a Software" paradigm powered by AI agents. Many have asked me: Why not bring this to Contact Centers? Why not build an AI-powered BPO?
Here’s why: the assumption of an "AI BPO" fundamentally misunderstands how the outsourced customer service (BPO) industry operates.
Let me explain.
In outsourcing functions like Legal and Accounting, customers only care about results—the what, not the how. They don’t concern themselves with the internal workings of the service provider: who they hire, what tools they use, or when they work. These are what I call fully outsourced functions.
But customer service and Contact Centers are a whole different ball game. BPOs in this space act as extensions of the customer’s team, not a separate black box. Clients care deeply about the how—where agents are hired, the software stack, training, policies, and more. These are what I call "extended team" functions.
Here’s the kicker:
For fully outsourced functions, introducing AI to improve efficiency is straightforward because the service provider owns the delivery model. Take Harvey, Haven - your modern tax advisor as an example—they can seamlessly increase their use of AI without consulting clients, as long as outcomes remain strong. Customers only care about results. So we will see a lot of new age Law Firms, Accounting firms, etc.
In contrast, for extended team functions, the decision to use AI rests with the customer, not the outsourcer. Clients dictate the blend of AI and human agents in their customer service operations—not the BPO.
So, can AI BPOs truly take off? Not until we fundamentally shift the way customers view their relationships with BPOs—and that’s a far taller order than most realize.