The two-sided future of customer service
The next battle in customer experience isn't human versus bot. It's the consumer's agent versus the brand's agent — and the layer in between is the whole opportunity.
Jesse Hollander · May 16, 2026 · 4 min read
Almost everything written about AI in customer service frames it as one contest: human agents versus AI agents. Will the bot replace the rep? How much of the contact center automates away? It's the wrong frame, and it's wrong in a way that hides where this is actually going.
The real shift is not that one side of the conversation gets automated. It's that both sides do. And once you see customer service as a conversation between two agents — one representing the customer, one representing the brand — almost every interesting question changes.
One side automated is just a faster status quo
Picture only the brand side automated. The company runs a capable AI agent; the customer is still a human, still navigating menus, still typing into a chat window, still doing the work of explaining themselves. That's the world most CX roadmaps are building. It is genuinely better than a phone tree. It is also just the existing relationship — brand serves customer, customer does the legwork — running faster.
The brand-side agent still answers to the brand. Its job description still includes deflecting contacts, protecting margin, steering toward the outcome the company prefers. A more capable agent on that side of the table is a more capable version of an institution whose interests are not the customer's. Better. Not different.
Two sides automated is a different relationship
Now put a real agent on the customer's side too — one that holds their whole portfolio of vendor relationships and acts only for them. The interaction stops being a person versus an institution's software. It becomes two agents, with opposed mandates, negotiating.
That is a structurally different thing. The consumer's agent does not get tired, does not miss the rate hike buried in the email, does not give up at the third menu, and does not forget what the brand promised in March. It carries the customer's history, preferences, and intent into every interaction — and it is loyal to exactly one party.
For the first time, the customer has representation that matches the institution's. Not better service. Symmetry.
We've made the long-form version of this argument before — the move from a one-to-one to a one-to-many topology, where one consumer agent transacts with many brand agents. The short version: the unit of customer service stops being "a company's support channel" and becomes "a negotiation between two represented parties."
The layer in between is the opportunity
Two agents with opposed mandates cannot just be pointed at each other. A negotiation needs a venue: a shared protocol, accepted identities, a way to produce a record both sides can rely on, a path to resolve disputes. Neither agent can be that venue — each works for one side. It has to be neutral.
That neutral layer is the actual prize. Not the consumer agent, not the brand agent — the connective tissue between them. It's the part that makes a two-sided interaction trustworthy instead of just two pieces of software talking past each other. We've called it exactly that: the connective tissue argument. It is the layer Teleperson is building toward, and it is the layer the whole thesis turns on.
What this means depending on where you sit
If you're a consumer: the thing worth having is not a better chatbot from each company you deal with. It's one agent that represents you across all of them. The value is in the representation and the portfolio, not in any single brand's tooling.
If you're a brand: the customer's agent is coming whether you build for it or not. The strategic question is whether you meet it as the easiest counterparty in your category or the most obstructive one. One of those compounds. The other churns.
If you're building in this space: the most defensible position is not either agent. It's the neutral layer — the marketplace, the protocol, the trust infrastructure — that both agents need and neither can be.
The human-versus-bot framing asks "how much of support gets automated." The two-sided framing asks a better question: once the customer finally has representation as capable as the institution's, what does the relationship between them become? Everything worth building over the next decade is an answer to that.