In this issue, we explore agentic commerce.  It has been making the rounds for the last few weeks, and I want to share with you what I found – and my perspective on it. 

First, my take:

I have said forever that commerce (or e-commerce, as we originally called it) was not supposed to be an application. From the very early Netscape eCommerce Server (remember? I do…) through the unnatural incarnations like Shopify – it has always been about glorifying a set of commoditized functions into something they shouldn't be. There is value in merchandising, pricing, shipping, inventory updates, order management, merchant services, payment gateways, etc., just not in one common, consolidated place where it: a) needs external providers to function, and b) those providers are far better quality than the consolidated functions within the commerce solutions themselves.

I have always advocated the use of common functionality from best-of-breed providers that connect into the enterprise architecture, not stand-alone solutions, and that are integrated deeply into the organization’s processes, not as a standalone server.  The true value of commerce is being able to fulfill the organization's standard sales processes with a different interface, without having to create a whole new set.

When it comes to Agentic AI, part of this advice has been taken: OpenAI integrated PayPal, Amazon, and others into their processes for one-stop shopping without changing channels. Amazon and Perplexity are battling over who owns the customer. Shopify tries to integrate frontier models into its operations. All this obviates the true value proposition: the ability to deliver agents that can be easily customized and that operate based on individual, customer, and contextual models, rather than the traditional, bloated, finite models integrated into the organization’s processes.

If you are interested in Agentic Commerce, please explore how you can get the functions you need integrated into your custom “world models” (I guess that’s the new name for customized models… sigh). And how those models adapt and integrate into your infrastructure. If you are looking for more details, here are some links I found:

1. iab (Interactive Advertising Bureau) released a study asking both consumers and providers for their opinions on Agentic Commerce. You need trade lead information for the study, but it’s a good collection of data that helps understand the choices and the work to be done. Quote: “AI doesn’t shorten the path to purchase — it reshapes it, creating new steps and opportunities” (which aligns with recent research that puts the value of AI for the organization in redesigned processes over the organization embracing AI.  Published October 2025.
   
   
2. McKinsey put out a white paper that dives deeper into the opportunity to mix agents and commerce, and what you can do (and shouldn’t do) about it. A good primer with lots of interesting data and insights. Published October 2025.
   
   
3. *Caveat: this is not a vendor recommendation* Former colleagues recently reached out to talk about what they are doing in Agentic Commerce. I was impressed with their approach: from FDW (forward-deployed engineers) to full advanced analytics, they target AI for retail with custom models, deep analytics, and hyper-contextualized solutions. Precisely what Agentic vendors should do instead of OOTB agents, check out Bayezon.ai and Quid Discover (in alphabetical order).