For three decades, outsourcing was sold the same way: by the seat, the hour, or the headcount. You paid for effort, whether or not the work was done well. AI is starting to put pressure on that logic, at least for the lower-effort, tier-one tickets where outcomes are now measurable and automatable enough that a small number of AI vendors can credibly price against them.
Whether that pricing model can extend to the kind of work most BPOs do is a different and much harder question. AI vendors can offer to charge per resolved ticket rather than per agent hour, because a missed resolution costs them a few cents of compute rather than a full day's wage. Almost no BPO can make the same bet today, since a missed resolution still costs a full agent’s salary regardless of outcome, and the operational risk of pricing that way doesn’t disappear.
What’s actually happening is more interesting than the headlines suggest, and this month’s reading gets into why.
What’s worth your attention this month?
1. Outsourcing providers built around labour have an incentive problem that no amount of AI messaging can paper over
Yuma AI's Guillaume Luccisano made a case worth mulling over this month on the CX Files podcast. Modern AI agents are no longer limited to answering questions from a knowledge base — they can access customer records, follow workflows, and execute transactions end to end. That capability shift is forcing a harder question for traditional providers than most are publicly acknowledging: Are they deploying AI to genuinely transform how they operate, or adding just enough of it to satisfy client demand while protecting a revenue model built on agent headcount? Luccisano's read is that the providers who survive this transition will look less like labour providers and more like systems integrators, selling expertise and orchestration rather than seats. The ones who struggle will be the ones whose business model depends on the thing AI is best at replacing.
2. The valuations behind AI customer service vendors are pricing something more specific than a better chatbot — and the gap to actually delivering it is wider than it looks
Victor Manzanera's analysis of recent funding rounds in the space is worth reading in full, but the core argument is this: companies like Sierra and Decagon are being valued at multiples that only make sense if investors believe they're capturing the labour budget itself, not just improving on existing software. The pricing logic behind that valuation — paying per resolved outcome rather than per seat — is real for those specific vendors, in the specific tier of work they target. It is genuinely difficult to extend the same model to a traditional BPO as the risk profile is completely different. That asymmetry is why outcome-based pricing remains rare even among providers who want to move toward it, and why the transitions that do happen tend to take the shape of a multi-month pilot with a single client rather than a wholesale pricing change. The complex, regulated, emotionally loaded work that is the actual core of outsourcing, still goes to a human, and that's where the more interesting shift is happening.
What do we think about it?
The complexity that AI can't resolve doesn't disappear — it lands somewhere, and increasingly it lands on whoever is left handling it
A pattern we're seeing repeatedly in conversations with CX leaders this month: the decision to outsource was rarely the problem; the partner behind it was. Teams end up handling the complex, business-critical cases themselves regardless of who's contracted to manage the broader operation, because the provider's model was built for volume rather than the judgment those cases require. As pricing in the industry shifts toward outcomes, that gap becomes harder to ignore. A provider optimised for headcount has little structural reason to build the kind of operation suited to the hard 20 percent that AI can't resolve cleanly, which is exactly the part of the work most worth designing for directly.
Where we’ve been
1. VivaTech 2026, Paris
Our Business Development Director, Kevin Mitchell, represented Otonomee at VivaTech this week, one of Europe's largest gatherings of AI and emerging technology companies. The energy in the room reflected the same shift running through this month's reading, a genuine reordering of how customer experience gets built and delivered, with no shortage of new entrants pushing the pace.
2. Fin Solution Partner Program — new hires
Following last month's partnership announcement, Otonomee is now hiring for two new roles to support our AI-enabled CX capability: an AI CX Product Builder who can move from problem to working prototype using AI tools, and a Full Stack Engineer to design and ship AI-powered applications end to end. Both roles are live on our website.
Closing thoughts
The question underneath this month's reading is really one of incentives. What AI is doing is forcing every provider to be honest about what their current model is optimised to deliver, and whether that matches what clients are starting to expect.
For CX leaders evaluating outsourcing partners right now, that's arguably the single most useful question to ask: What is this provider's business optimised to protect, and does that match what you need from them?
If you're reassessing how your support operation is structured — or simply want to compare outcomes against your current provider before making any changes — it's a conversation worth having early.
.jpg)
