Lessons on Agentic Commerce: Chapter 2

As I covered in one of my last posts, agentic commerce is mostly a two-part system: agents for buyers and agents (or agent ready storefronts) for sellers, across both B2C and B2B environments.

Where was chapter 1? That is something I have been building and refining as a showcase for the last 90 days, in my free time, at HomeGadgets. Chapter 1 is primarily data acquisition at scale: current, from legitimate sources, and accurate about what a given market actually offers. Today I have collected and indexed a real-time database of 17,000+ unique SKUs across more than 80 retailers and over 100 brands. And growing. At the moment is far from “perfect”, however is well over 85% accurate, with more work needed to reach 99.999%.

Why the focus on legitimacy, accuracy, depth and up to date information?

Because that is the remedy against what I call “lazy AI” that is filled with incomplete and/or out-of-date answers that will inevitably drive a shopping agent to make “sub-optimal” work (to say it euphemistically).

So, what is chapter 2?

Enter the LLM — currently living at the /ai subdomain. Rather than describe it, go try it: https://www.homegadgets.ca/ai. In this release you can ask questions with real specificity:

What is the cheapest and most energy efficient <52dB dishwasher available to ship in Vancouver?

Where can I shop for the cheapest GE refrigerator model GNE25DYRKFS near the Greater Toronto Area?

And so on.

Later iterations will include additional context that further advance a purchase decision: delivery areas, fees, tax, haul-away fees, delivery windows, and the rest.

Why does this matter?

Your favorite AI chatbot or search engine will most likely fail at this level of specificity. Having deep, current data is the moat. It is what lets a shopping agent representing a consumer answer with accuracy, legitimacy, depth and truly immune to the “pay to play” world of online advertising (see example below)

The implications are not small. Among other things, this guts a major subsegment of online advertising and the SEO budget merchants pour into being found. Merchants now care about being found by the robot and being able to interact with it as it now rewards quality data. That alone deserves its own deep dive: how zero-click visits, and bots rather than people arriving at your site, are reshaping the web.

The next chapter is the third of four on the consumer end of agentic shopping: the MCP server. After that is the agentic payment layer and I will then close the loop with the “B” side of B2C.

Happy shopping!