Transforming business with AI

About ten years ago, when I had one of the world's largest retailers as a client, I visited their headquarters building for a meeting with some of their senior executives. Entering the building, in order to enter the main office area, one had to walk down a corridor where on either side you could see room after room of vendors pitching their products to the company's buyers.

Sitting at the bar at the hotel that night I met one such vendor who described his experience. He had driven with his sample products in his car into town in order to have his one hour face to face meeting with the retailer's buyer. For him the meeting was a make-or-break opportunity for his business. The volume, if he earned the business, might crush his small company but without the distribution this retailer could provide he would also fail.

I don't know how things turned out for him but I did come away from the experience convinced that, while a necessary part of the retailer's operations, this was a terrible process both for the buyers and the sellers. Couldn't anything be done to improve how this retailer screened potential products for their stores? It has been 10 years since then so perhaps things have changed a lot. Or maybe not.

But what I do believe is that this is the kind of business process that can be radically transformed using the current, or at least a coming soon, generation of artificial intelligence. The logic that buyers are applying to decide on what to purchase can be codified. Vendors could be pitching to an AI buyer that can evaluate the product and the vendor. All of this could be done with less anxiety and effort by all parties.

Let's take the process apart step by step and think about what transformation might look like.

Phase 1: Introducing a mock interview for sellers

As a first step the retailer might create an online AI agent which serves as a training/testing experience for the sellers. Just as in the physical presence of a buyer, the seller would pitch their product to the AI agent. But in this first phase the AI agent would simply provide (in a helpful and supportive way) an evaluation of the presentation with tips to the seller about how to improve but also guidance on what the buyer is likely to like or dislike about the product, value proposition, or terms. For example, perhaps the seller neglects to say anything about their ability to deliver certain quantities of the product in given time frames. The bot might provide a prompt saying "you'll need to inform the buyer that you can produce and deliver at least X quantity every Y period for this type of product."

The seller can then refine and improve the pitch, the product, and the offer before ever getting in front of a human buyer - making both buyer and seller's time more effective when they do meet.

Phase 2: Screening for meeting requests

As a second step the retailer could utilize a refined and improved version of the first phase AI agent to recommend approving or rejecting vendor requests for meetings. As before the vendor would pitch to the AI agent and the AI agent would provide feedback. But in addition the AI agent would prepare a summary of the interaction and a recommendation to the buyer as to whether or not the buyer should schedule a meeting. The decision would still remain in the hands of a human - responsibility for decisions should always be in human hands with the duty of care being higher as the consequences of the decision increase. But substantial amounts of everyone's time can be saved if vendors who are unlikely to move forward in the process can be screened out in this way.

Phase 3: Making buying recommendations

Eventually improvements in the AI agent can lead to full screening and recommendations for product purchases. An AI agent can patiently take as much time as is needed, and as many follow-up meetings as is needed, with a seller to work out details on pricing, production, and specific product configurations. A complete dossier on the opportunity would then be automatically prepared and delivered to the human buyer - again putting the ultimate decision into human hands. But by this time the relationship of buyer and seller, and their respective roles and activities have vastly changed. Buyers are now able to spend quality time with just those vendors that have reached the point of providing maximum value to the retailer. Vendors now have the chance to pitch multiple times, refining their offer until it makes sense for both parties to move forward or ceasing engagement if there isn't a business value to be found.

This isn't that far away. Take a look at some of the newest capabilities for analyzing video streams in Anthropic's Claude 3.5 or OpenAI's 4o and you can see where this is going. Massive reinvention of every business process is beginning now and will continue for the next decade. What are you reinventing?