Skip to main content
Back to News
news/AI Applications

Amazon Turns Alexa Into a Shopping Agent You Can Text

Amazon is testing Alexa for Shopping over text messages, letting invited US customers search, compare and buy by SMS ahead of the holiday season.

Stefan Trbojevic

Stefan Trbojevic

6 November 20262 min read
LinkedIn
Frosted glass cubes on a dark slate surface linked by copper wires with warm amber light inside

The takeaway

The interface was never the hard part. Amazon is building a first-party text agent and blocking outside ones, so agent commerce will be negotiated per merchant, authorization-first, not scraped from open APIs.

Why it matters for builders

Text-based agent commerce puts authorization in the flow: consent, spend caps and price triggers must be encoded because there is no checkout screen. Expect closed storefronts and per-merchant access deals, plus phone-number identity as the new linking layer.

Amazon Turns Alexa Into a Shopping Agent You Can Text

Amazon started texting its own customers on Friday. A small, invitation-only group of U.S. shoppers received a message inviting them to try Alexa for Shopping over SMS: accept the terms, link a phone number, and the assistant will search products, compare options, build a cart, complete the purchase, and follow up with delivery updates. No app, no website.

Amazon says the invite will expand to millions more customers in the coming weeks, and shoppers who do not get one can join a waitlist. The contact shows up in the phone as "Alexa Shopping."

What Amazon actually shipped

The feature extends an assistant that already lives in the Amazon Shopping app and on the web. Amazon says it can search hundreds of millions of products across its store and the wider web, check up to a full year of price history, track deals, and text a customer when an item falls to a target price. It accepts photos: send a picture of an outfit and ask for similar looks. It remembers stated preferences, so a vegan-only request or marathon training sticks for future recommendations.

"Customers often jot down what they need the moment they think of it, in their notes, in their messages, in their group chats," Rajiv Mehta, Amazon's vice president of Conversational Shopping, told The Verge. The pitch is that the shopping list already lives in a text thread.

Abstract render of glowing amber nodes linked by light filaments on a dark surface

Why the channel matters more than the feature

Texting is where consumer agent traffic is heading, and Amazon is arriving with closed doors behind it. The company blocked Meta's Muse agent from querying its site in September, then launched its own text-based buyer weeks before the holiday shopping season.

The competitive evidence is thin but pointed. DoorDash said last month that third-party agents had placed only hundreds of orders in total on its platform, and that it would concentrate on its own text agent instead. Merchants are not handing over checkout to outside agents. They are building first-party ones and writing the rules for everyone else.

What this means for builders

If your product touches commerce, the constraint is no longer model quality, it is authorization. A text-based agent completes a purchase with no visual confirmation, so consent, spend caps, and price triggers have to be encoded in the flow rather than assumed from a checkout screen. Phone-number linking also makes the handset the identity layer for shopping agents, which brings its own consent and deliverability problems.

The other lesson is architectural: expect closed storefronts. Agent-commerce access will be negotiated per merchant, not taken from a public API. The realistic build is a first-party agent with a thin, reliable text surface, not a universal shopping bot roaming the web. The early agent-commerce numbers already suggested consumer volume is small. Amazon just proved the interface is not the bottleneck. Access is.

Share𝕏

The Automation Brief

Read 5 AI stories instead of 50.

The essential moves in AI agents, models, automation and infrastructure — filtered for builders and operators, with the part that actually matters.

No noise. Unsubscribe anytime.

Editorial notes

Reported by

Stefan Trbojevic

Edited by

n8n Lab Editorial

Published

6 November 2026

Updated

6 November 2026

AI disclosure: AI assisted with research and drafting. Factual claims are reviewed by an editor.

n8n Lab is an independent service provider. We are not affiliated with, endorsed by, or sponsored by n8n GmbH. “n8n” is a trademark of n8n GmbH and is used here only to describe the platform-specific implementation and automation services we provide.