How technology changes commerce.
I research AI Commerce, digital transformation, and the systems shaping how products are discovered, evaluated, and bought.
I've spent more than 20 years working across ecommerce, merchandising, marketplaces, Amazon, digital commerce, and technology. This site is where I document what I'm learning about how AI and other technological shifts are changing commerce, and what operators need to understand before the shift becomes obvious.
What I'm exploring
The questions I keep returning to as technology reshapes how commerce works, and how the pieces connect across disciplines.
AI Commerce
How AI is reshaping the way products get discovered, evaluated, and bought across marketplaces and retail. Rufus, Sparky, ChatGPT, and the agentic surfaces are the current front line.
Product Discovery
How shoppers, and increasingly their agents, actually find products. What surfaces a listing, what gets skipped, and how discovery is shifting from search boxes to conversations.
Digital Transformation
How commerce operations, tooling, and data systems evolve when technology moves faster than the org chart. What it takes for teams to keep up without simply adding headcount.
Search, GEO, and AI Recommendations
How search is moving from keyword ranking to AI recommendation. Generative Engine Optimization, Agentic Engine Optimization, and what decides whether an engine cites or recommends you.
AI Commerce 2027: From Experiment to Operating Model
A living outlook on what 2026 actually built, what it enables next, and what remains unproven. Updated through 2027 as evidence arrives.
Recent research
Google Now Explains Your Brand to People Who Already Know It
In late September, AI Overviews spread across branded Google searches, from 5 of 98 bare brand names on July 1 to 63 of 100 on September 30 in Ahrefs' panel. Branded search did not overtake non-branded; it caught up. The open question is whether the answer sits above the brand's own result, and what it says when it does.
The Hardest Part of Agentic Commerce May Be Knowing What the Shopper Wants
Agentic commerce has two representation problems. Whether the agent understands the product is the one the industry has worked on. Anthropic's controlled book market found the other one, whether the agent understands the person, accounted for 85% of the gap from the best possible outcome, and that a full model upgrade moved human outcomes by 0.01.
Amazon Just Put Channel Mix Inside the Algorithm
In March I wrote that budget allocation, target ROAS, channel mix and product prioritization should stay with a human. Amazon's September 29 Full-Funnel Campaigns documentation puts two of those four inside the optimizer, and the DVA+ capability table shows Amazon now selling delegation level as a product choice.
This is not an AI news site. It is a public notebook for understanding how technology changes commerce. The goal is to compound knowledge, connect ideas across disciplines, and document the questions, experiments, and patterns that help operators make better decisions.