September 27, 2026
TechCrunch reported that Google is testing a feature letting users in India buy products from Walmart-owned Flipkart directly through Gemini and AI Mode, with a limited rollout now and a broader launch planned for October. On its face this looks like a small regional pilot. It isn't. It's Google testing whether an AI assistant can replace the entire shopping funnel -- search, compare, decide, pay -- inside a single conversation. If that works in India, a market where Flipkart already competes head-to-head with Amazon, expect it to travel fast to other markets and other retail partners.
The business implication is bigger than e-commerce. Any company whose product depends on being 'found' -- through search ads, marketplace placement, or a branded app -- should be asking who does the finding when an AI agent does the shopping instead of a human clicking through ten tabs. If Gemini becomes the default shopping interface, the businesses that win are the ones with clean, structured product and pricing data that an agent can actually parse. That's the same discipline behind a well-run CRM: the businesses whose customer and product data are organized and current are the ones AI tools can actually act on. The ones with messy spreadsheets get skipped, quietly, by an algorithm that never has to explain itself.
While Google chases the transaction inside a chat window, Meta is chasing it through hardware you wear. TechCrunch's coverage of Meta Connect described smart glasses as the dominant theme of the event, with Meta clearly positioning them as the successor to the phone as the primary way people stay connected. Separately, Meta opened early access to new features for Muse, its AI agent-building tool, distributed in a very on-brand way: you have to ask Muse itself to get on the list.
These two moves aren't unrelated. Glasses are the delivery device; Muse-style agents are what runs on top of them. Meta's long game is an AI layer that sits on your face and acts on your behalf, the same ambition Google has for its chat interface and the same ambition every vendor selling you an 'AI agent' has for your workflow. The lesson for any business evaluating these tools isn't which company wins the glasses war -- it's that the agent doing the acting needs to be one you control, with clear rules about what it can see and do. That's the argument for building on workflow automation you own rather than bolting a third-party agent onto your stack and hoping it behaves. We've made a version of this point before about who really owns the agent when a platform vendor builds it for you.
A smaller but telling story: a TechCrunch writer built an interactive digital avatar of themselves, trained it on their own expertise, and came away genuinely conflicted about what it means to let an AI clone speak in your voice, on your authority, without you in the room. That ambivalence is worth sitting with. Every trend above -- Gemini shopping, Meta's agents, wearable AI -- depends on trusting an automated system to represent you accurately. A shopping agent misreading your intent is annoying. A cloned professional voice giving wrong advice under your name is a liability problem. Businesses experimenting with AI avatars or agents for sales and support should treat identity and accountability as a design requirement, not an afterthought.
Which of these is the bigger deal to you: an AI that can finish your shopping cart, or one that can wear your face while it talks to strangers?
Sources