Gen AI has taken the world by a storm for the past 3 years. It promises increased productivity on a personal and professional level. I believe nearly every job or trade has been affected by it in some way. Some are affected more significantly than others. I am not debating whether AI will replace us humans in our current jobs.

I believe AI will take over tasks that are repetitive and time consuming. This change will allow us to focus on strategy and creative aspects of the role. In theory, this should lead to better role satisfaction. Practically, how organizations go about in implementing the AI transformation will decide how it impacts its employees roles and experience.

From personal initiatives to enterprise tools

Many professionals have taken it upon themselves to use gen AI tools publicly available (ChatGPT, Gemini, Claude, Copilot etc.) to enhance their productivity at work. However, corporate policies around data security, intellectual property, and confidentiality usually intervene quickly by blocking employees from utilizing these publicly available tools in the workplace.

Suppliers of e-procurement tools such as SAP, Ivalua, Coupa, Zycus have also augmented their existing suite of enteprice procurement systems with Large Language Models helpers, however not all companies utilizing these systems have enabled these additions for either cost concerns, data security issues, or not havig enough clarity internally on what is permissible and what is not so “disable it until further input” is the safe decision.

A third opportunity (though limited in scale) presents itself to software providers like Microsoft with it Co-pilot and Google with Gemini. These suppliers are already embedded in their customers’ systems either through their operating systems and other office applications that can easily be integrated with their AI offering.

Finally, some buyers have started developing their own LLM in house or used an existing model and customized it for their own private use. This offers a higher level of security for their information and intellectual prototype by isolating their own LLM environment from the public, but obviously requires a much higher technical know how and investments in infrastructure and systems.

Use cases examples

When we look at how procurement and sourcing professionals can leverage Gen AI in their work, there are a few common themes:

• Market Research
  Gen AI, with the proper prompting or with an agentic workflow, can be used to do the following:
  • Identify key companies in a specific product/service category, identify competitors to familiar incumbents either onshore or offshore and create a more competitive roster of bidders.
  • Extract data on public companies for supplier market research that allows the procurement professional to evaluate the financially viable companies to consider further.
  • Identify contact persons for new supplier companies to establish the first discussions.
    
• RFx drafting:
  Gen AI can be used to draft RFx requirements. This is only recommended if you have fine-tuned or provided Augmentation of the LLM (through RAG) of your organization procurement policies, your standard RFx templates and conditions, your list of requirements.
  
• Negotiations preparation:
  AN interesting application of Gen AI is that it can be used to practice your important negotiations sessions. With the proper prompting of a LLM you can assign a negotiator persona, objectives, motivations to it as a negotiator stakeholder and set the context and history, which would then allow you to probe it for key arguments and counter-arguments on different positions, or even in a more creative mode to discover additional opportunities.
  
  With an agentic workflow you can setup multiple such negotiator personas to reflect actual negotiations scenarios.
  While this is not a substitute for human social cues in actual negotiations I think it is a step over the traditional negotiation paper based preparation that uncover some blind spots.
  
• Paralegal support:
  For professionals that have to work on contracts lacking legal training, Gen AI can be a life saver. It can be asked to:
  • Interpret existing legal clauses (legalese is a tricky language to master)
  • Propose a proper legal language around an idea explained in plain human language.
  • Review a contract and allow the user to ask questions about it and chat with the AI around the content.
    
• RFx evaluation:
  • Defining evaluation criteria for the RFx based on specific inputs required and objectives.
  • A clear evaluation matrix with clear criteria can be used to automate the evaluation of supplier responses.
  • Offer summaries of RFx results and generate reports in specified templates.
    
• Contract implementation:
  • Agentic workflows to turn supplier quotes sent by the budget owner to verify contract compliance and populate requisition for approval process.
  • Agentic workflow to verify service levels against a commitments and calculate service credits and report performance for supplier management.

Human in the loop

For obvious reasons, all these potential use cases will still require a sanity check from a human professional. LLM workflows allow for a human in the loop to make sure that any key decision is rationalized by the right level of responsibility.

Considerations

Organizations should think holistically about their AI transformation journey, and avoid departments building their own silos. The raw materials an organization brings in this journey is its data, their need to be a data confidentiality, policy, cleanup, utilization/monetization, and bias avoidance strategy. This will help much of the following steps to be done in a logical manner. By attacking the sticky topic of what is our data like and how can we leverage it and what needs to be done to do so, will avoid a lot of turning around in circles later on.

Feedback needed

Thank you for reading so far, if you have experimented with Gen AI, LLM or traditional Machine learning (regression, classifiers, re-enforcement learning) in procurement, I’d be very interested to know more, please leave a comment.

Acronyms and definitions

Gen AI: Generative Artifical Intelligence. Popularized by Large Language Models (LLM) such as ChatGPT, Claude etc. Gen AI excels in interpreting natural language and creating content.

LLM: Large Language Models, a type of artificial intelligence program that is designed to understand, generate, and process human language. They are called “large” because they are trained on massive datasets of text and code, often containing billions or even trillions of words and sentences.

Agentic AI: refers to a type of artificial intelligence system that is designed to act autonomously, make decisions, and achieve complex goals with limited human supervision. The term “agentic” highlights its capacity for independent and purposeful action, or “agency.”

RAG: Retrieval-Augmented Generation (RAG) is an artificial intelligence (AI) framework designed to enhance the capabilities of Large Language Models (LLMs) by providing them with access to external, up-to-date, and domain-specific information.

RFx: short for Request for Quotation (RFQ), Request for Proposal (RFP) and so on. It is basically a sourcing event where a buyer specifies requirements and conditions and invites suppliers to respond and provide proposals against this requirements. Reverse auctions or electronic auctions are also a type of RFx events in this context.