AI Commerce Fragmentation Requires New Monetization Infrastructure

We've built the technology. We're solving the integrations. But there's one question nobody has fully answered: Do customers trust AI agents to shop for them? 🛡️ Trust is the ultimate currency of agentic commerce. If a customer doesn't trust the agent to get the right size, the right price, or the right delivery window, they'll go back to browsing manually. Building trust isn't a technical problem; it's a brand problem. It requires transparency, reliability, and a clear "human escalation" path. How are you building trust in your AI-powered customer interactions? 🎤 #AITrust #AgenticCommerce #RetailAI #GoogleCloud #CustomerExperience #BrandStrategy

Extremely impressive demo today with Richard Ward and Daksh Gupta of StackCX. Before the demo, having done some research, I genuinely thought the platform might be too detailed for a dealer group of our size. How wrong I was. It’s exactly the opposite. The level of visibility, control and insight it can provide is exactly what I’ve wanted for years, and, on more than one occasion, probably gone to bed dreaming about! For me, the exciting part is the transparency. Being able to understand what is really happening across customer conversations, measure performance consistently and identify opportunities to improve the customer experience has huge potential. AI is moving incredibly quickly, but seeing a practical application that can genuinely change how we operate and manage our businesses is something else entirely. Now waiting rather impatiently for our go-live date… Very excited to see the results I already know this application will deliver across all departments of Alan Day Motor Group.

The homepage shouldn’t control every buyer journey. Someone landing on a $1.2M listing from Google has already supplied context: the property. I’m thinking about Auren Web as page-aware sales infrastructure, not one generic chatbot across an entire brokerage site. The page should change where the conversation begins. Good AI doesn’t make every visitor start from zero.

The follow-up gap is where businesses die. Most sellers get one message, don't get an immediate 'yes', and move on. The reality is that most sales occur between message 2 and message 4. If you aren't automating that follow-up, you are leaving thousands on the table. Let our AI do the work. Claim your 500 free credits at WACC360.app. #LeadNurturing #SalesStrategy #Efficiency

The follow-up gap is where businesses die. Most sellers get one message, don't get an immediate 'yes', and move on. The reality is that most sales occur between message 2 and message 4. If you aren't automating that follow-up, you are leaving thousands on the table. Let our AI do the work. Claim your 500 free credits at WACC360.app. #LeadNurturing #SalesStrategy #Efficiency

The follow-up gap is where businesses die. Most sellers get one message, don't get an immediate 'yes', and move on. The reality is that most sales occur between message 2 and message 4. If you aren't automating that follow-up, you are leaving thousands on the table. Let our AI do the work. Claim your 100 free credits at WACC360.app. #LeadNurturing #SalesStrategy #Efficiency

Google put a lip-syncing face on Gemini 3.8 Live agents, in 97 languages, for Gemini Enterprise. The first reviews called it creepy. The avatar is a visual layer on the September 15 Gemini 3.8 Live models, Digital Trends reports. The agent underneath still triggers tool calls and fetches data in the background while the face talks. A refund happens in a tool call whether or not a customer is looking at a generated person. None of the coverage includes customer data. Engadget and TechRadar both put "creepy" in their headlines, and no outlet reports a pilot result or a price for the avatar layer. The one cost figure, Google's via TechRadar, is that Live Extended Thinking runs cheaper per hour of input audio than two rival voice models. Disclosure sits with the deployer. Google built its SynthID watermark to be imperceptible, so a customer learns the agent is software only if the company says so on screen. A face built on a brand's own likeness needs Google's enterprise allowlisting; the first test runs on one of Google's presets. Start with voice alone, add the face only in flows where it shows the customer something, and score the two on repeat contacts within seven days and handoffs to a human. Which of your support flows would a face improve: a walkthrough, or a billing dispute? #customerservice #gemini #voiceagents Read the full story: https://lnkd.in/eJ3VTMUQ

The other day I wrote about AI agents starting to show up in real estate companies’ sales funnels. Here’s the other side of that. A web agent handling a full storage rental, including PCI-compliant payment. That matters for two reasons: lower cost of sale for the operator, and a much easier experience for the customer. This is the part I’m especially excited about. Fully autonomous agents that can actually complete the transaction, not just talk about it. #AIAgents #Storage

Today we released Claude Commerce Agents: an open-source blueprint for building a shopping agent inside your own app and a merchant agent for the teams who run the store. Working code, four industry examples, and a Claude Code plugin that fits it to your catalog and systems. It runs where you already build: Claude API, Bedrock, Vertex AI, Foundry. If your team is planning what ships before peak season, this is a fast way to start. Learn more below or reach out to me if your team is interested in building agents for commerce! https://lnkd.in/g_Mj8Ri6

Anthropic just released Claude Commerce Agents, an open-source blueprint for businesses to build commerce agents inside their own apps. It is designed for merchant-owned experiences rather than a new Claude shopping marketplace. The blueprint includes a shopping agent that can search products, compare options, answer policy questions and build a cart. It also includes a merchant agent that can analyze performance, maintain listings, monitor inventory and draft pricing or promotional changes. Checkout remains with the merchant, and operational changes require human approval. This joins two other major moves already happening in agentic commerce. OpenAI and Shopify connected Shopify Catalog to ChatGPT through the Agentic Commerce Protocol. Shopify product data can appear during conversations in ChatGPT, with customers completing purchases through the merchant’s own checkout. Google and Shopify also co-developed the Universal Commerce Protocol. UCP gives agents, retailers and payment providers a shared way to support discovery, purchasing and post-purchase activity across Google Search, Gemini and other platforms. Agentic commerce is now developing on two fronts. External AI platforms are becoming new product discovery channels. At the same time, businesses can build their own agents to help customers shop once they arrive. Both depend on the same thing. Structured product data. An external agent needs clear data before it can find, compare and surface a product. A merchant’s own agent needs that data before it can answer questions or make a useful recommendation. The same listing now has to serve customers, search engines, answer engines and shopping agents. SEO remains important for getting the product discovered. Meaningful AEO makes the listing clear enough for an agent to understand who it is for, how it compares and whether it fits the request. That means clean product titles, schema, categories, attributes, variant relationships, dimensions, materials, compatibility, price, inventory, policies, FAQs, reviews and current availability. A product can rank well and still be useless to an agent if the listing cannot answer the customer’s actual question. This is why SEO and AEO can no longer sit apart from catalog management and commerce data. The brands that treat their listings as structured knowledge will give agents the clearest reason to find, understand and recommend their products. Everyone else may technically be available while never making the answer.