AI everywhere: which prerequisites does AI in procurement need – and how do you check?
Five prerequisites for AI in procurement – use case, data, systems, skills and rules – plus a self-assessment that shows in about an hour where your team stands.
This article is also available in Deutsch.
AI in procurement rests on five prerequisites: a clear use case, usable data, the right systems and tools, a team with basic AI skills, and binding rules. You find out whether they are in place with a structured assessment – the self-assessment below gives you a first indication in about an hour.
AI is everywhere – so why are the basics still missing?
AI has reached the mainstream. In Bitkom’s latest survey of 603 German companies with 20 or more employees, 57 per cent use AI, up from 36 per cent a year earlier. Yet only 3 per cent say they are exploiting much of its potential (Bitkom, 14 Sept 2026, in German).
The bottlenecks are rarely the models – they are the foundations. Companies not yet using AI cite a lack of technical know-how (85 per cent), legal uncertainty (66 per cent) and costs that are hard to estimate (63 per cent); 40 per cent lack the necessary data. Among companies already using AI, data protection (66 per cent) and legal uncertainty (56 per cent) top the list. The survey names infrastructure (51 per cent) and data preparation (50 per cent) as the biggest cost items (Bitkom, 14 Sept 2026).
For procurement, the question is not whether AI is coming, but whether your team has the prerequisites to make AI in procurement work day to day.
Which five prerequisites does AI in procurement need?
| Basic | What it means in procurement | Signs that it is missing |
|---|---|---|
| 1. Goal and use case | A specific task with a measurable goal – such as classifying spend, comparing quotes or pre-screening suppliers. | “We want to do something with AI”, but nobody can say which task will be faster or better afterwards. |
| 2. Usable data | Order, invoice, supplier, contract or catalogue data that is complete enough for this use case and can be found. | Data is scattered across emails, spreadsheets and folders; categories and supplier names are inconsistent. |
| 3. Systems and tools | Access to the data (export or interface from the ERP or e-procurement system) and an AI tool approved by the company. | Staff use private AI accounts because there is no approved tool – or nobody can get at the data. |
| 4. People and skills | The team knows what AI can and cannot do, can write good instructions and checks results critically. | AI is either not used at all, or its output is accepted without checking. |
| 5. Rules and accountability | Clear guidance: which data may go into which tool, who approves, who is responsible for the result. | Uncertainty about data protection and trade secrets slows everything down – or there are no rules at all. |
The order matters: the use case determines which data, systems and rules you need. Teams that first try to “fix the data” without knowing the use case risk cleaning up in the wrong place.
Three typical starting points – and what each one requires
Three applications that many procurement teams start with show how much the prerequisites differ:
- Classifying spend: AI assigns order and invoice lines to categories. It needs line data with description, supplier and amount, plus a category structure to map against. The tool must handle larger data volumes, and someone in procurement has to check samples.
- Comparing quotes: AI extracts quotes and lines up prices, terms and deviations. It needs the request as a reference, quotes in digital form and a clear rule on which documents may go into which tool – quotes often contain confidential terms.
- Pre-screening suppliers: AI summarises publicly available information on a supplier. It needs clear screening criteria and a team that can judge sources and results critically – AI answers can be wrong or out of date.
The three examples show that not every use case needs the same data. Starting with a task whose prerequisites are largely in place makes the benefit visible sooner.
How do you find out where your procurement team stands?
Take an hour with two or three people from procurement and IT and answer the following questions honestly with yes or no. Keep one specific use case in mind – not “AI in general”.
How to read the result:
- Mostly yes: you can start with a clearly scoped pilot. Define success criteria up front.
- No mainly on 1–2: collect and assess use cases first – a workshop with the procurement team is a good way to do this.
- No mainly on 3–7: check which data your most important use case really needs and sort out access. With a clearly defined scope, this step stays manageable.
- No mainly on 8–12: before the technology, you need training and rules. Without them, any tool will go unused or be used in risky ways.
The self-assessment does not replace an analysis, but it shows where your first step lies.
Four steps from question to first project
- Assess: run the self-assessment for two or three obvious use cases and note the gaps.
- Prioritise use cases: weigh benefit against effort realistically. Start where data exists and the benefit becomes visible quickly.
- Close gaps selectively: add only the data, access, training and rules that the prioritised use case needs.
- Define the pilot: agree goal, timeframe, success criteria and owners in writing – and only then choose a tool.
What does the EU AI Act require on AI literacy?
Article 4 of the AI Act on AI literacy has applied since 2 February 2025 (Regulation (EU) 2024/1689). Regulation (EU) 2026/1744 of 8 July 2026, published in the Official Journal on 24 July 2026 and in force since 27 July 2026, replaced the article: providers and deployers of AI systems take measures to support the development of AI literacy of their staff and of other persons dealing with AI systems on their behalf. The text states explicitly that they are not required to guarantee a specific level of AI literacy for any individual (Official Journal of the EU, L 2026/1744).
In practice, the task for procurement stays the same: whoever introduces AI tools needs training. Documenting it is advisable – the law does not prescribe a particular form. According to Bitkom, 70 per cent of companies already offer AI training, yet two thirds rate their employees’ AI skills as low (Bitkom, 14 Sept 2026).
Your next step
If you would rather not do the assessment alone: in the AI readiness check we go through the five prerequisites for AI in procurement with your team in 2–3 hours. The result is a prioritised list of use cases, the main gaps in data and systems, and a recommendation for the first step. If your use case is already defined, vendor-neutral AI tool selection is the right starting point; an obvious first use case is AI-supported spend analysis.
For a no-obligation initial call, just use the contact form.
Sources
- Bitkom e. V.: Erstmals nutzt die Mehrheit der Unternehmen KI (in German), press release of 14 Sept 2026 (survey of 603 companies with 20+ employees, calendar weeks 28–33/2026)
- Regulation (EU) 2024/1689 (AI Act), EUR-Lex
- Regulation (EU) 2026/1744 of 8 July 2026 (Digital Omnibus on AI), Official Journal of the EU, 24 July 2026
- IHK (German Chamber of Commerce): Digital-Omnibus-Verordnung zur KI tritt in Kraft (in German)
Frequently asked questions
What are the minimum prerequisites for AI in procurement?
Five basics: a clearly defined use case with a goal, usable data on orders, suppliers or contracts, access to the systems and to an approved AI tool, a team with basic AI skills, and binding rules on data protection and confidentiality.
Does our data have to be perfect before we start with AI?
No. The data has to be good enough for the chosen use case, not for every conceivable one. That is why it pays to define the use case first and then check exactly the data it needs.
How do I find out whether our procurement function is ready for AI?
With a structured assessment of goals, data, systems, skills and rules for each use case. The self-assessment in this article gives you a first indication; an AI readiness check turns it into a prioritised list of use cases in 2–3 hours.
Do procurement staff need AI training?
The EU AI Act requires organisations that deploy AI systems to take measures to support the AI literacy of their staff (Article 4). Since the amendment by Regulation (EU) 2026/1744, this is an obligation to take measures to support AI literacy – nobody has to guarantee a specific level. Either way, training is what makes AI useful in day-to-day work.