AI as a labour service bypasses the adoption problem
A delivery model in which vendors sell finished work product, not AI tools, removes internal-adoption friction from the buyer side and accelerates displacement timelines.
Most of the existing analysis of AI disruption assumes the buyer takes on the adoption problem. A firm buys AI tools, trains its staff, redesigns its workflows, and eventually captures the value — or, more often, does not (see A tools-first AI rollout that plateaued). Adoption friction is what has been protecting incumbents from the full pace of the technology, because each internal transformation is slow, expensive and partial.
An emerging delivery model removes the buyer’s adoption problem altogether. The vendor provides the AI, the prompts, the quality assurance, and the domain expertise. The client provides the data. The vendor delivers finished work product — reports, analyses, regulatory filings, client communications — under commercial terms. The buyer does not adopt AI; the buyer buys an outcome.
Why this changes the displacement calculus
The friction that internal adoption imposes is real but one-sided. It slows the rate at which incumbents can absorb the technology, but it offers no protection against competitors who have already absorbed it and are selling through. A firm that could not make AI work internally is at no less risk from a vendor who can.
Three things accelerate when the delivery model exists. First, the customer’s decision compresses to price and quality, both of which the vendor is well-positioned to optimise. Second, the vendor underwrites the risk of AI failure, which is exactly what has made internal adoption expensive; removing that risk from the buyer is substantial value. Third, the vendor accumulates cross-client data and prompt tuning that the buyer cannot match internally, widening the capability gap over time.
What it implies for incumbents
The framing “we will adopt AI at our own pace” is a framing for the internal transformation problem, not for the external displacement problem. Internal-adoption friction is no protection against external disruption captures the inference. Firms whose competitive position depends on the codifiable layer of their work (see AI commoditises general expertise) face a different question than has been widely acknowledged: not “how fast can we adopt AI?” but “how fast can an outside vendor deliver the work we currently sell?”.
For professional services in particular, the risk vector is not the firm’s own tooling programme. It is the AI-native provider who will arrive with a finished deliverable at a price the incumbent cannot match.