5 top generative AI use cases in procurement

5 use cases for generative AI in procurement

Five practical applications where GenAI could be potentially useful to a procurement team include the following.

1.      Supplier messages and RFP development

GenAI can draft text, such as supplier emails, RFPs and agreements, by drawing on information from past documents and organizational templates. The output still needs human review and refinement, but generative AI can provide a human employee with a draft to work from.

2.      Contract review and summarization

Reading, extracting information from and comparing terms between hundreds of supplier agreements takes up a large portion of a procurement manager's day. Generative AI can scan contracts and identify non-standard clauses, flag potential compliance issues and summarize any risks.

GenAI often does well at identifying patterns across document sets, such as audit reports or renewal terms, across multiple vendors. The technology can identify whether any reports are concerning or whether agreements lack important text, such as indemnification clauses.

3.      Spend analysis and insights

Conventional spend analytics require line items to be categorized manually and aggregated. Generative AI does not require that degree of organization to analyze spending patterns, identify anomalies and, most usefully, generate natural language explanations of any anomalies.

GenAI's capabilities can be particularly useful for tail spend--the small-value purchases that individually don't justify the procurement team's attention but add up to a significant proportion of expenses. For example, generative AI can draft a message to tail spend suppliers.

4.      Supplier risk assessment

A diligent procurement team must monitor the financial health of suppliers, along with their operational capacity and their history of regulatory compliance. These processes typically happen only periodically--sometimes just annually--instead of continuously.

GenAI can gather and synthesize supplier risk data through automated analysis, market intelligence and performance metrics. The technology can process information such as news feeds, financial reports and regulatory filings to assess the stability of suppliers, then recommend action if needed.

Generative AI can also monitor mid-tier and smaller suppliers, which normally don't receive dedicated relationship management.

5.      Purchase requisition support

Requisitions that fall outside standard procurement workflows can create bottlenecks and compliance issues. An AI assistant can advise workers with unusual requests about supplier selection and can check that purchases align with existing contracts and policies.

GenAI can also send messages to suppliers to confirm specifications, ask for quotes and negotiate basic terms.

The benefits of using generative AI for procurement

Generative AI can help procurement teams carry out tasks that they otherwise wouldn't have time to do. For example, it enables procurement workers to analyze and optimize categories that often didn't receive attention because of their modest spend.

GenAI can also help procurement employees improve their business decisions. For example, a human employee can prepare for a negotiation by having generative AI tool analyze historical pricing, market data, suppliers' financial information and macroeconomic trends. A human employee would potentially only gather that amount of data for a major contract, but GenAI can synthesize that information prior to all negotiations.

The risks of using generative AI for procurement

However, chief procurement officers might be, understandably, concerned that AI-generated outputs could contain errors or that the technology could misinterpret business context. AI hallucinations that occur in the preparation of a commercial contract could have serious consequences.

Data privacy and intellectual property, such as proprietary supplier information, also require safeguards. Generative AI use in procurement must include the proper access controls and audit trails.

Retrieval-augmented generation (RAG) architectures can help address several governance concerns by ensuring that GenAI responses are based on organizational documents instead of relying solely on the training data for a model. For example, if a procurement manager asks about contract terms or supplier analysis, RAG retrieves the information from a secure internal document repository instead of drawing on public training sets.

A RAG architecture helps ensure that proprietary information is kept secure and that audit trails are available, as users can trace each recommendation back to specific source documents.

RAG doesn't eliminate the need for data quality management or human review. However, it substantially reduces the risks of hallucination and intellectual property leakage.

How to begin implementing generative AI for procurement

Procurement teams should begin using GenAI by launching well-scoped pilots in areas where mistakes will result in limited consequences.

Contract summarization and spend analysis are good potential starting points because human employees can easily check the outputs against source documents.

The team should establish KPIs before implementation and build governance frameworks that define quality and compliance standards before employees start using the technology.

Most important is treating early generative AI deployment as a learning exercise. The technology's output will likely improve, but the procurement team must develop the processes and judgment to deploy it effectively. Careful experimentation is a good approach.

Donald Farmer is a data strategist with 30-plus years of experience, including as a product team leader at Microsoft and Qlik. He advises global clients on data, analytics, AI and innovation strategy, with expertise spanning from tech giants to startups. He lives in an experimental woodland home near Seattle.