For an industrial services company, we utilized generative AI (Gemini) in supplier market analysis for recycling certain equipment. Here’s our return on experience based on this work.
Time-Saving, Like Having a Junior Assistant
Generative AI undeniably accelerated our research. No more spending hours compiling scattered data: a few well-formulated queries are enough to get relevant summaries on our potential suppliers. AI acts like a junior assistant who helps you on demand, requires precise instructions, doesn’t always bring back gold nuggets, but significantly helps you move forward. That said, AI does not replace in-depth manual research for complex or poorly documented cases on the Web.
Personalized Analyses, Provided You Guide the AI Well
Generative AI has the potential to provide customized analyses, but you have to learn how to question it effectively. Precise and contextualized queries are essential to get relevant answers. Sometimes you have to rephrase, refine, and iterate to get the desired result. It is often necessary to broaden the questions so that the AI brings back several answers that you will select instead of it. Indeed, when the AI sees several possible answers, it tends to say that it doesn’t know. It’s an interactive tool, not a crystal ball.
New Information, To Be Taken With a Grain of Salt
Generative AI can highlight interesting trends or weak signals, but its analytical capabilities are limited by the data it has access to. It is therefore always necessary to cross-check the information, compare it with other sources, and use our own business expertise. In particular, there are quite a few consistency errors in the figures that can confuse the AI (for example, the tonnage collected can be interchanged with the number of pieces of equipment collected). Hence the need for additional questions to cross-check the figures or even requests on the different versions available of an indicator to decide for yourself. AI is a decision support tool, but you have to help it not to decide for you.
Potential to Exploit, With Pragmatism
Generative AI is promising for supplier market analysis, but it is not yet the miracle solution. It requires an investment in time to master it, and its results must be interpreted with discernment. Used pragmatically, it is a valuable ally to gain efficiency and make more informed decisions.
Some Tips and Tricks for Querying the AI
To get the most relevant information from the AI, it is crucial to formulate clear and precise queries. Here are a few tips:
- Be specific: Avoid vague and general questions. Specify the type of information you are looking for (financial data, reputation, risks, etc.) and the context of your request (industry, geographical area, etc.).
- Use relevant keywords: Identify the key terms related to your search and include them in your query. This will help the AI understand your needs and provide more targeted results.
- Structure your query: Organize your request in a logical way, using short and simple sentences. If necessary, divide your query into several sub-questions to facilitate the AI’s understanding.
- Don’t ask it for a closed answer about data, rather ask it for all the available variants of that data and choose the right one.
- Don’t talk to it formally, you introduce « noise » into the queries and sometimes it even confuses the result with the process to get the result: « Can you give this… » reports a process to find what is requested, « Give this… » reports what is requested
- Be patient: The AI may need some time to process your request and provide relevant results. If you are not satisfied with the first results, rephrase your question or refine your search criteria.
Generative AI is an interesting tool to integrate into our buyer’s toolbox, but it does not replace our judgment or experience. It is by using it in a complementary way to our skills that we learn to get the most out of it and to save time and gain depth in our supplier market analyses.
Pssss – this article, based on my true experience, was built with help from Gemini as you can imagine…
