The AI-First Procurement Team

Everyone’s talking about AI replacing procurement jobs. Almost nobody’s showing what it actually looks like when a team is deploying agents, using foundation models, and treating AI as a genuine working partner.

Not predictions.

Not theory.

Here’s the work and how it’s changing.

The Gap Between Talk and Reality

There’s no shortage of articles about how AI will transform procurement.

Most of them read like they were written by people who’ve never actually used the tools they’re describing.

Vague promises about ‘streamlining workflows’ and ‘unlocking strategic value.’

Here’s the reality: The tech is here.

Some teams are using it. Most are watching.

I’m going to show you what’s actually happening with demos, not descriptions.

But I’m also going to be honest about what we don’t know yet, because anyone claiming certainty about where this is heading is either lying or selling something.

What we CAN’T predict:

• Exactly what job titles will exist in 18 months

• What org charts will look like

• Which specific roles will grow or shrink

What we CAN show:

• The work that’s already being done by humans and AI together

• How contracts get reviewed

• How vendors get onboarded

• How analysis gets done when AI is genuinely embedded

Let’s start with what’s disappearing.

The Work That’s Disappearing

(And What Replaces It)

If you’re a category manager or buyer today, here’s an honest assessment of how you probably spend your time:

About 80% of it is busy work.

Chasing documents. Formatting spreadsheets. Copying data between systems.

Writing the same emails to different suppliers.

Reviewing the same clauses in different contracts.

Updating the same trackers every week.

Building the same reports every month.

None of this is difficult. It’s just... time-consuming. And tedious. And endless.

This work doesn’t get ‘transformed’ by AI. It vanishes.

The contract review that took two hours?

An agent does the first pass in three minutes.

The supplier research that consumed an entire afternoon?

Synthesised while you make coffee.

The data cleanup that nobody wanted to do?

AI extraction tools handle it trivially.

So what fills the space?

Honest answer: We don’t entirely know yet. That’s not evasion—it’s the reality of being in the middle of a transition this significant.

What we ARE seeing:

More time on supplier relationships. The human parts. The conversations that build trust. The negotiations that require reading the room. When AI handles the prep work, you show up to meetings better prepared and with more energy for what actually matters.

More time on strategic questions. The judgment parts. Should we consolidate suppliers here? Is this risk acceptable given our context? What’s the right trade-off between cost and resilience? These questions require human context that AI doesn’t have.

More time figuring out how to use AI better. Meta-work. Designing workflows. Evaluating new tools. Training colleagues. This is temporary—eventually AI collaboration becomes unconscious competence—but right now it’s a real part of the job.

The uncomfortable truth: We don’t yet know what ‘full-time procurement professional’ looks like when 80% of current tasks are automated. The work is changing faster than job descriptions are updating.

Agents Doing the Busy Work

Enough theory.

Let me show you what this actually looks like.

The following demos aren’t mockups or promotional videos.

They’re actual recordings of agents doing actual work. Watch what happens, then notice what the human still needs to do.

But first—

If ‘AI agent’ still feels vague to you, start here.

I made a 20-minute video breaking down what an agent actually is, how it differs from ChatGPT, and why it matters specifically for procurement. It covers the perceive → reason → act → learn framework with a live demo. Everything in this section will make more sense after watching it.

20 minutes. No jargon. Includes a live Gatekeeper agent demo.

But if you’re serious, we put together a mega page at Gatekeeper. An AI Agent Library full of agentic use cases. This doesn’t cover every use case for Procurement. We cover Contract and Third Party Management.

But it will get the juices flowing.

Check it out here by clicking this button

AI Agent Library

What We Don’t Know Yet

Will there be fewer procurement jobs?

Probably in some organisations.

Definitely different jobs in all organisations.

The honest answer is: it depends on whether the work that expands (relationships, strategy, judgment) requires as many people as the work that contracts (data handling, chasing, routine analysis).

Some teams will get smaller.

Some will stay the same size but do fundamentally different work.

Some will grow because they can now do things that weren’t previously possible.

What will job titles look like?

No idea.

Anyone claiming certainty is selling something.

‘Category Manager’ might still exist, but the work will be unrecognisable.

New titles will emerge for roles that don’t exist yet—Agent Orchestrators, maybe, or AI Workflow Designers.

Job titles always lag reality by years.

They did during the last technology transitions too.

What We ARE Seeing

Patterns from teams that have actually deployed AI

People who understand both procurement AND AI are becoming invaluable.

Not AI specialists.

Not procurement people who’ve taken a prompt engineering course.

People who genuinely understand both domains well enough to bridge them.

They’re rare, and they’re shaping how this transition unfolds in their organisations.

The boring parts of every role are being automated first.

This is consistent across every team I’ve talked to.

The first things to go are the tasks nobody liked doing anyway.

The work that felt like bureaucratic overhead rather than meaningful contribution.

Relationship and judgment work is expanding to fill the gap.

When you’re not spending four hours on data prep, you have four hours for other things.

Most of that is going toward work that requires human capabilities—conversations, decisions, strategy.

Teams are smaller but need higher-skilled people.

This isn’t universal yet, but it’s a pattern.

Ten people doing routine work get replaced by five people doing judgment work.

The five remaining roles require more capability and pay more.

The implications for hiring and development are significant.

The Takeaway: Focus on the Work, Not the Titles

Job titles will lag reality by years.

They always do. When email emerged, we didn’t immediately create ‘Email Coordinators’—we just started using email as part of existing roles until the work evolved enough that new titles made sense.

The same thing is happening now.

The work is changing faster than the org charts.

So what should you actually do?

Stop asking ‘what role should I aim for?’

That framing assumes someone has a map of where this is going.

Nobody does.

The roles that exist in 18 months haven’t been defined yet.

Start asking ‘what work am I doing that AI can handle, and what work requires me?’

Make a list. A

ctually write it down.

You probably have tasks in both categories and clarity about which is which will serve you better than any job title prediction.

Develop the capabilities that remain human.

Systems thinking.

Behavioural economics.

Ethical judgment.

Relationship building.

These don’t become less valuable as AI handles more.

They become more concentrated.

Learn to work with AI, not just alongside it.

Position yourself for whatever emerges.

The teams that figure this out first will have an 18-month head start.

The individuals who position themselves well will have leverage regardless of how the org charts eventually settle.

The work is changing now.

Your title can catch up later.

The AI Procurement Blueprint is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.