Buying Outcomes, Not Just Labor

HR and procurement no longer buy only labor. They buy an outcome that might come from a combination of people and AI.¹ Some talent firms now design models that include both. One is rPotential, a spin-off of the staffing group Adecco.

In our reading, buying outcomes changes where the playbook starts, and the seven actions follow from it.

1. Map tasks and outcomes

Start each sourcing discussion by breaking a role or project into its tasks and the outcomes they produce. High-volume data validation or repetitive call-center functions might go to AI agents. Tasks that need complex judgment, persuasion, or deep domain expertise may stay with people or use a hybrid approach.

In our view, the buyer cannot price an outcome until the work is split this way. Without the breakdown, procurement falls back to buying headcount or software seats.

2. Catalog what AI can do

Build an internal taxonomy that maps AI capabilities to common roles. Examples include marketing analyst, customer-support rep, and scheduling coordinator. Language models might excel at drafting marketing copy. Specialized computer-vision agents might be best for quality checks in manufacturing.

The catalog is procurement’s request for proposals (RFP) for AI. When the buyer knows which models solve which problems, it can ask vendors to prove domain expertise instead of paying for hype.

3. Draw the handoffs

An outcome delivered by people and agents together needs clear role boundaries. A hybrid-workforce strategy defines which tasks AI owns, which tasks people own, and how problems escalate.

For example, an AI customer-service agent might receive a complaint about a complex billing dispute. If the amount is above a set dollar threshold, a rule might route it to a human specialist. Documented roles and handoff points prevent conflict and duplicated work.

4. Redesign the workforce model

The labor mix now includes full-time employees, temporary hires, freelancers, and AI. Models to consider include:

1. Client-owned digital labor.
   The company licenses or builds its own AI and brings “digital employees” in-house.
2. Leased digital labor.
   The company rents AI agents from a third party, like traditional temp staffing.
3. Fully outsourced AI subdepartments.
   A vendor runs an entire process, such as order fulfillment or a call center, with AI and a small team of human experts.

Leasing AI might fit seasonal spikes. Full outsourcing could be more efficient for repetitive but critical functions. We looked at these options in how to buy AI agents.

Managing the mix requires new key performance indicators (KPIs) and cost structures. They reflect digital labor’s economics: scalability, near-constant uptime, and quick retraining.

5. Set legal and ethical ground rules

In our view, a buyer that pays for an outcome still carries the risk of how it was produced. HR and procurement work with legal, compliance, and ethics teams to draft enterprise-wide standards for AI use. Those policies cover:

1. Proprietary data.
   Whether and how AI learns from the company’s data.
2. Bias.
   How to detect and remedy it.
3. Privacy.
   How to safeguard personal or sensitive information.

A global company faces a patchwork of regulations across jurisdictions. Firms that build these frameworks early can adjust to new AI legislation better than firms that wait until after a law passes.

6. Capture value as it evolves

Staffing has traditionally followed a “set it and forget it” approach. That approach doesn’t fit a human-AI team. AI’s value compounds as it learns from interactions.

Feedback loops track performance, update training data, and revise sourcing strategies. If an AI scheduling tool keeps hitting edge cases, it may need more training or more human oversight.

The contract should let the buyer capture improvements over time. It should also respect the vendor’s intellectual property (IP) and the company’s data sovereignty.

7. Invest in the people

As AI takes on mundane tasks, high-value work led by people matters more. Competitors can’t simply download that work. Training helps employees work alongside AI and use it to amplify their own impact.

Capabilities to build include relationship building, ethical decision-making, and creativity.

• • •

Where buying outcomes gets hard

In our reading, buying an outcome is simpler to describe than to contract for. The seven actions leave the hardest ownership questions open.

Say an agent learns from the company’s proprietary data while it delivers an outcome. Who owns the resulting capability is not yet settled: the company, or the owner of the model. If the agent makes a mistake, who bears liability is also open. So is whether agents need employment-like contracts.

Nobody needs every answer upfront. The open questions work as a guide, answered along the way. In our reading, the outcome gives HR and procurement a place to start. The open questions show where the contracts still need work.