Skip to main content
Contact

AI how-to guides Finance

How to use AI in procurement

Short answer

Bring purchase orders, invoices and contracts together, let AI classify the spend and point out savings, supplier issues and upcoming renewals, and automate the routine transactions. Buyers keep the negotiation, the supplier relationship and the final sourcing decision. Start with spend visibility in a few categories, because every other use depends on it.

Updated: · Thinkia

What AI does

  • Consolidates spend from ERP, invoices, purchasing cards and contracts and classifies it by category, supplier and business unit.
  • Spots consolidation opportunities, off-contract buying and contract leakage.
  • Scores suppliers continuously on delivery, quality and compliance data.
  • Flags upcoming contract renewals with performance and savings analysis attached.
  • Drafts purchase orders, routes approvals and handles supplier messages for standard transactions.

What stays with people

  • Category strategy and the final sourcing decision.
  • Negotiating with suppliers and managing the relationship.
  • Approving purchases above thresholds or outside policy.
  • Judging supplier risk where the data is thin or the context matters.

How to put it in place

  1. Get one view of spend

    Extract purchase orders, invoices and contract data from the systems where they live. Normalise supplier names and map spend to a category taxonomy, either your own or a standard one such as UNSPSC.

  2. Pick the categories with the most room

    Choose a few categories with significant spend, many suppliers or frequent off-contract buying. Agree with the category managers which questions they need answered and what decisions the analysis should support.

  3. Let AI find opportunities and buyers check them

    Run classification, duplicate supplier detection and leakage analysis, and have buyers review the findings. Their corrections improve the classification and show which signals are worth tracking.

  4. Add supplier scoring and renewal alerts

    Combine delivery, quality and compliance data per supplier and attach it to each upcoming renewal. Buyers then go into the negotiation with the facts, not only memory.

  5. Automate standard transactions

    For low-risk, catalogue or repeat purchases, let an agent draft the order, route approval and message the supplier. Keep approval thresholds and policies in the systems, not in the AI's instructions.

  6. Measure and extend to more categories

    Track the quality of classification, findings accepted by buyers, cycle time and compliance with contracts. Extend to new categories with the same playbook once those hold.

What you need

  • Access to ERP, purchasing, invoice and contract data, including purchasing cards if used.
  • A category taxonomy and a supplier master that someone owns.
  • Supplier performance data: deliveries, quality incidents, compliance checks.
  • Category managers willing to review findings and correct the classification.
  • Clear approval thresholds and purchasing policies the automation must follow.

When it is not worth it

  • Low spend concentrated in a few suppliers that the team already knows well.
  • When purchase data is mostly on paper or spread across spreadsheets nobody can access.
  • Public tenders where the procedure fixes most of the choices: focus on document preparation instead.

Risks and how to control them

  • Wrong classification leading to wrong conclusions: review samples with buyers and show the data behind each finding.
  • Messy supplier master data: deduplicate and fix it first, or savings and scores will be misleading.
  • Automation bypassing controls: enforce thresholds and segregation of duties in the systems, and log every automated action.
  • Confidential pricing and supplier data: restrict access by role and do not send it to tools without contractual guarantees.
  • Unfair or opaque supplier scoring: make criteria explicit and let buyers see and challenge the inputs.

How Thinkia does it

We start with a spend lens: categories, suppliers and contract coverage are baselined, and quick wins are separated from strategic bets. Signals and workflows such as RFx assistance, should-cost models and risk flags then connect to your purchase-to-pay and source-to-pay tools.

Buyers run real events with AI support in a category pilot while savings, cycle time and compliance are measured. Playbooks and templates then extend to more categories, with supplier onboarding and ESG checks following the same rails.

The Thinkia solution behind this guide

Thinkia products involved

Frequently asked questions

What can AI do in procurement?

Classify and consolidate spend, find savings and off-contract buying, score suppliers, flag renewals, help prepare tenders and automate routine purchase orders and approvals. The decisions on strategy and negotiation stay with buyers.

Do we need clean data before starting?

Clean enough to trust the main categories and suppliers. AI helps with classification and deduplication, but the supplier master and taxonomy need an owner. Starting with a few categories makes the cleanup manageable.

Can AI negotiate with suppliers?

It can prepare the negotiation with prices, performance and alternatives, and draft messages. The negotiation itself, and any commitment, should stay with a person.

Is this different from a spend analytics tool?

Spend analytics shows where money goes. AI adds automatic classification of messy descriptions, detection of opportunities and risks, and agents that act on routine transactions, all connected to your purchasing workflow.

Where should we start?

With one view of spend for a few categories. Visibility is the base for savings, supplier scoring and automation, and it shows quickly where the data needs fixing.

Related guides

Related decisions

Key terms

Sources

Want to do this in your organisation? Let's talk.

Talk to an AI Expert