Acquire Intelligence: What Acquire BPO Became, and What Its AI Arm Actually Does

The automation and intelligence line of business, announced in February 2024, spans four named capability areas:

• Robotic process automation. Rule-based automation of repetitive back office steps.
• Chatbots. Conversational handling of routine customer contacts.
• Voice biometrics. Caller verification by voiceprint rather than knowledge-based questions.
• Speech analytics. Analysis of call audio for quality, compliance and insight.

That is a recognisable and fairly conventional stack. It is worth noting what it is not: this is operational automation layered onto outsourced delivery, not a general-purpose AI platform sold standalone. For buyers that is often the more useful proposition, because the automation arrives attached to people who run the process.

Nearly every provider in the market now describes itself as AI-enabled, which makes the label close to meaningless as a selection criterion. These questions separate deployed capability from positioning, and they apply to any provider, not just this one:

• What is in production, and for how many clients? A named reference running the capability live is worth more than a demo.
• Who owns the automation once it is built? If you fund the build, establish whether you can take it with you.
• Where does data go? Which models process your customer data, in which jurisdiction, and under what retention terms.
• Where is the human in the loop? Specifically, what happens when the model is wrong, and who is accountable for the outcome.
• How does the commercial model change? If automation removes handling time, ask whether you keep the savings or the provider does. A per-agent price and an automation program pull in opposite directions.
• What is the accuracy baseline? Measured on your data, not on a vendor benchmark.

The last two questions are the ones that most often go unasked, and they are where the value actually sits.

Automation and per-agent pricing pull in opposite directions, and this is where AI-enabled outsourcing contracts most often go wrong. If you pay per agent per month and the provider deploys automation that removes a third of the handling volume, the saving accrues to whoever wrote the contract more carefully.

Three structures are common. Per-agent pricing is simple and familiar, but it gives the provider no incentive to automate away its own revenue. Per-transaction or per-outcome pricing aligns the incentives better and is harder to specify. Hybrid arrangements, where a baseline is fixed and automation savings are shared on an agreed split, are increasingly the practical middle ground. Decide which one you are buying before you evaluate the AI capability, because the capability is worth very different amounts under each.