Procurement 5.0: how people and AI share the work

In July 2025, McKinsey described Procurement 5.0 as five pillars: digital and AI adoption, supply-chain resilience, supplier partnerships, ESG, and changes to the operating model. The European Commission's 2021 Industry 5.0 report sets related priorities for industry: people, sustainability and resilience.

For a procurement team, the question is what changes in the next tender. Who writes the requirements? Who checks the bids? What information does the person making the decision receive?

What Procurement 5.0 means

McKinsey frames the term around procurement's role across the enterprise. Lynkaz puts human and machine collaboration among its central elements: software handles repetitive, data-heavy work while people concentrate on relationships and decisions.

Our working interpretation starts with that division of responsibility. AI can assemble information and check documents against defined requirements. Procurement professionals set the objectives, examine the findings and take responsibility for the outcome. The wider Procurement 5.0 agenda also asks them to consider supplier resilience and environmental or social objectives.

The term is useful when it helps a team specify those responsibilities. Calling a system "5.0" tells the buyer very little about which tasks it performs or how its work can be checked.

Follow a tender through the workflow

Consider an illustrative purchase of equipment that needs installation and ongoing maintenance. The buyer has a technical specification, a delivery schedule and a draft contract. Suppliers need to understand the package well enough to price the work and submit the required evidence.

At preparation, we would give AI a defined review task: extract the requirements and look for contradictions across the documents. A delivery date in the specification might conflict with the installation schedule. The specialist would review both passages, establish which date is intended and correct the package before publication. Choosing a feasible deadline still requires knowledge of the project.

Market research has a similar handoff. An agent can search for candidate suppliers and collect evidence of relevant products or previous contracts. The buyer then assesses whether those examples fit the purchase. A supplier's experience with similar equipment is useful information; its capacity to deliver this contract may need a direct conversation.

On the supplier side, the same requirements can become a preparation checklist. The bid team can use it to track the requested certificates and responses, with a person responsible for confirming that each item is complete. An automatically drafted answer still needs someone who knows what the company can deliver.

During evaluation, AI can compare the submitted material with the requirements and flag a possible gap. Imagine a maintenance commitment that appears in the technical response but is absent from the pricing schedule. The useful output would show the requirement and both relevant passages. The evaluator can then investigate the discrepancy and determine its significance under the tender's rules.

Data needs a source and a scope

The OECD's review of AI in public procurement describes applications including spend analysis and supplier scouting. It also identifies skewed data and limited explainability as risks. A procurement team needs to understand the information behind a result before acting on it.

For the equipment purchase, a previous contract value needs context: what equipment it covered, whether installation was included, and when the contract was awarded. We would keep those details beside any price comparison. Otherwise, the buyer has to reconstruct the basis for the estimate before using it.

Document findings need the same discipline. We would expect a reviewer to be able to open the cited passage, see which requirement it relates to and record a correction. If the system cannot establish something from the available documents, the uncertainty should remain visible. A source citation gives the reviewer somewhere to start; they must still check that the passage supports the conclusion.

Sustainability and resilience need explicit requirements

The broader frameworks include responsibilities that document automation alone cannot fulfil. McKinsey discusses supplier collaboration and sustainability alongside AI. The Commission's Industry 5.0 vision puts worker wellbeing and environmental limits within the purpose of technology adoption.

For our example purchase, we would translate those objectives into questions the buyer can investigate. Can the equipment be repaired? How will spare parts be supplied? What evidence supports the supplier's energy-use claims? The buyer needs to decide which considerations belong in the specification and how to assess them.

AI can help locate the relevant commitments in a bid. Measuring performance after delivery requires its own data and follow-up. Any team adopting this approach should define who owns that work as well as who runs the initial analysis.

Where Tendergate fits

We are building Tendergate around these handoffs. Its bid analysis checks proposals against tender requirements and connects findings to source evidence. The specialist reviews what the agents found and decides what to do with it.

In market research, the user approves the items to research and reviews the sourced supplier and price information. Tender publishing brings supplier invitations, questions and amendments into the same workspace as the tender. These cover part of the Procurement 5.0 agenda; the team's wider supplier strategy and sustainability work remain separate responsibilities.

To assess whether this division works for your team, choose a completed tender and compare the AI's findings with your recorded review. Inspect the source passages. Record missed issues, incorrect flags and the time needed to check the output. Use that exercise to decide which tasks to delegate and where a human review is required.