What is the role of procurement in the new reality of AI-enabled services? (Updated: May 2026)

Not just an intelligent typewriter.

When this post was first published in January 2023, ChatGPT had been publicly available for six weeks. Most people were still treating it as a curiosity - a clever text generator that could write a cover letter or summarise a document. I argued at the time that procurement leaders were missing the bigger picture. Three years on, I think that case is considerably stronger, and the window for action has narrowed.

So let me update the argument.

Where we were in 2023

Back then, large language models were impressive but limited. GPT-3 was the prevailing architecture, hallucinations were frequent, and the practical applications in a business context required significant human oversight. The question was whether the technology would mature fast enough to matter. It has, and then some.

Today, we are operating in a world of GPT-4 and its successors, multimodal models that process text, images, and structured data simultaneously, and purpose-built AI agents capable of executing multi-step business workflows autonomously. The question is no longer whether AI will reshape the service supply base. It already is.

The procurement logic, restated

The fundamental reason companies outsource services to suppliers has always been straightforward: suppliers can perform certain tasks more efficiently, more quickly, and at lower cost than an internal team could. That calculus is now being disrupted at scale.

Marketing copy, contract drafting, RFP document preparation, supplier risk screening, spend analysis, customer service scripting, code review, translation - these were all classic candidates for external service providers. AI-enabled internal teams can now perform many of these tasks at a fraction of the cost and time. The question procurement leaders should be asking is not whether to use AI, but which parts of their external service spend can legitimately be brought back in-house.

This is not theoretical. Companies that have moved deliberately on this - building internal AI capability rather than simply buying AI-enabled services from the same suppliers who provided manual services before - are seeing material reductions in external spend alongside improvements in speed and output quality.

What procurement needs to do

The starting point is a structured review of the service supply base through an AI lens. For each category of external service spend, the questions are: what is the core task being performed, how much of that task is now automatable with current AI tools, and what residual human judgement is genuinely required?

Some categories will hold. Complex litigation, regulatory advisory, highly contextual strategic consulting, and bespoke engineering work still require experienced external specialists. But a substantial portion of the service supply base in most companies - particularly in areas like communications, documentation, compliance checking, and routine analytics - is now fair game for insourcing.

The second step is the internal conversation. Procurement rarely controls service category decisions unilaterally. Building the business case and engaging internal customers - marketing, legal, finance, HR - on the cost and quality opportunity is where the real work lies. That conversation is easier now that the technology has matured enough to demonstrate rather than merely describe.

The risk of waiting

Companies that moved late on low-cost country sourcing in the 1990s and 2000s did not disappear overnight. But they ceded competitive ground that took years to recover. The dynamics with AI are faster. The organisations building internal AI capability today are compressing cost structures and response times in ways that are difficult to replicate quickly from a standing start.

Procurement has always been at its best when it sees around corners. This is one of those moments.

Stay safe. Be bold.

Daniel

The views expressed in this post are my personal professional opinions, based on research and publicly available information. They reflect analysis of industry trends and practices, not assertions of fact about specific companies or individuals. Nothing in this post constitutes legal, financial, or investment advice.