It’s 11pm, and a customer wonders whether a particular jacket comes in a larger size. She doesn’t open a browser, search for the shop, scroll through a catalogue and hunt for a contact form. She sends a message. Within seconds she has an answer, a recommendation for a similar style, and a way to pay without leaving the conversation. Nobody on the shop’s staff was awake.
That scene is still partly aspirational, but this autumn it moved noticeably closer to ordinary. Two things are happening at once: businesses are starting to sell inside chat apps, and AI assistants are starting to shop on behalf of customers. Each changes the familiar path of search, click, browse, add to cart and fill in a form, and together they make that path look a lot less fixed.
In this post we’ll look at what Meta, Walmart and others announced, why trust has become the hard question, what the shift means for Indian businesses in particular, and what you can do now to be ready. We’ll finish with a checklist and answers to common questions.
Two Shifts at Once
The first shift is businesses selling through conversations. A customer asks a question on WhatsApp, Messenger or Instagram, gets a recommendation and, increasingly, can complete the purchase in the same thread. The second shift runs the other way: AI agents browsing, comparing and buying for people, which we covered in our guide to agentic commerce.
For businesses, the practical consequence is the same. More of the sales funnel is moving to places where you don’t control the page, and where the customer, or their assistant, expects clear answers fast.
The Numbers Behind the Shift
Meta says more than one billion active conversations take place every day between people and businesses on WhatsApp, Messenger and Instagram, and more than one million businesses already use versions of its Business Agent.
Until recently, most of those chats ended with a link to a website. Meta’s aim is for more of them to end with a sale. When it launched Business Agent globally on June 3, 2026, at its Conversations event in London, it described an agent that answers questions, recommends products, books appointments and hands difficult queries to a person. The tool had been tested for nearly two years in markets including India and Mexico.
Meta describes its test as letting people buy “right after learning about it from a brand’s agent.”
Meta Moves Checkout Into the Conversation
On October 6, 2026, at Advertising Week in New York, Meta announced a test of direct checkout inside Meta Business Agent on Messenger, working with Shopify, so a person can buy a product without leaving the chat. It’s also testing checkout in Meta AI with Adyen, PayPal, Shopify and Stripe.
It’s important to be precise about status. These are tests, not a general rollout, and availability will vary by market and account. The same day, Meta previewed a Personal Agent Protocol, an open standard it’s building with Sierra to govern how personal AI agents deal with businesses, with partners including Shopify, Stripe and Walmart. Early specifications were expected later in October.
For shops that run on Shopify, the practical advice from the platform commentary is simple: treat your product feed as shopper-facing copy. If you want help preparing that side of a store, our web and e-commerce team works on exactly this layer.
Walmart Opens the Door to AI Shoppers
If Meta shows businesses selling in chat, Walmart shows the other half. On October 21, 2026, it announced three tools. The Walmart Agent Gateway lets approved AI assistants, including ChatGPT, Gemini and Meta’s Muse, connect to Walmart and Sam’s Club systems to search products and complete purchases. It starts as a beta for companies and developers and uses the Model Context Protocol, a standard for connecting AI applications to outside systems. Agent Profiles, due in November, give each authorised agent its own profile inside a customer’s account, so the agent’s orders appear separately from the customer’s own. And a new set of terms spells out what agents may do, with rule-breakers blocked.
The controls are the interesting part. Shoppers can require approval for each purchase by default or set a spending limit, with $50, $100 and $200 given as examples, and they can switch a profile off or delete it. Walmart has said it wants shoppers to choose the apps and agents they prefer, while keeping control.
The contrast with Amazon is striking. From September 21, Amazon blocked Meta’s Muse agent from shopping its site, saying an unauthorised AI agent violates its conditions of use. One major retailer has built a front door for agents, the other has locked it, and which approach wins will shape how people buy for years.
The Trust Question: Banks, Credentials and Control
None of this works without trust, and this month showed why. Axios reported on October 22 that major US banks, including JPMorgan, Bank of America, Morgan Stanley and Wells Fargo, have detected unauthorised third-party AI agents operating inside customer accounts, and Reuters independently confirmed the reports. In most cases, according to the coverage, the agent logged in with the customer’s own credentials. Banks have been moving to block or restrict such access, and Citi said it doesn’t permit third-party AI agents to log in or transact within its digital banking platforms.
The lesson for retail is the same one Walmart’s design reflects: an agent shouldn’t borrow a customer’s password. It should have its own identity, its own limits and a clear record of what it did. Chat-based selling carries older risks too, including fake business accounts, phishing links and payment scams, so verified business profiles and payments through trusted rails matter more as more money moves through conversations.
What This Means for Indian Businesses
India has been running one of the world’s largest conversational commerce experiments for years. Business Standard estimates the country accounts for more than 850 million WhatsApp users, close to a third of the platform’s global audience, and the country has more than 87 million micro, small and medium enterprises. Many small sellers already take orders by message. In May, Meta launched Business AI for businesses in India, with support in Indian languages, 24/7 replies, lead capture, appointment booking and product recommendations, and said UPI payments inside the chat would follow. It cited a 2025 Kantar study finding that 91% of online adults in India chat with businesses weekly.
One caution on payments: WhatsApp Pay runs on UPI but remains a small player next to PhonePe and Google Pay, so many businesses still share payment links or QR codes from other UPI apps. The opportunity is real, and the rollout of in-chat payments is still maturing. For first-time online shoppers outside the big cities, a familiar chat app often feels safer than an unfamiliar website, which is why this shift may matter most there.
How to Prepare: A Practical Plan
Treat chat as a sales channel, not a support inbox. Set up a verified business account, keep a complete catalogue and decide how payment will work inside the conversation. Meta reports more than one billion active conversations between people and businesses every day, so your customers are already in chat, and the question is whether yours is set up to sell to them.
Get your product data in order. Agents can only recommend what they can understand. Clear names, accurate prices, stock levels, sizes, delivery estimates and return policies make you easier to find. It’s the same kind of structured detail we built into the product pages for Gutta på Haugen, and it’s what agents need as well as people. The more complete the information, the less an agent or a shopper has to guess, and guessing is where recommendations and sales get lost.
Keep a human in the loop. Let an agent handle routine questions at midnight, but route complaints, complex orders and high-value sales to a person, with a smooth handover. A clear handover also leaves a record of what the agent has already said, so the person taking over doesn’t have to start from scratch. We build that kind of setup through Kilowott Intelligence.
Automate the routine, not the relationship. Order updates, stock replies and abandoned-basket reminders are good candidates, which is where automation helps. Start with a small set of repeat questions and expand once you’ve seen how customers respond.
Make your ads lead to the conversation. Click-to-message ads can start a chat directly, and our digital marketing team can help connect campaigns to the conversation that follows. Measuring those chats separately from website visits also shows you which campaigns actually start conversations that end in a sale.
Fix the basics first. If your standard checkout is slow or confusing, chat won’t repair it, as we explain in our guide to mobile checkout. A chat-to-checkout path is only as strong as its weakest step, so test the whole journey on a phone before you promote it.
A Conversational Commerce Readiness Checklist
- Verified business account set up on the chat apps your customers use
- Complete, accurate product catalogue, with names, prices, sizes and stock
- Delivery times, costs and return policy written clearly and kept current
- In-chat or link-based payment option tested end to end
- Clear rules for when a person takes over from the agent
- Only trusted payment methods used, with customers told what to expect
- Approach decided for AI agents that visit your store, including limits and terms
- Product feed treated as customer-facing copy and reviewed regularly
- Chat conversions tracked separately from website sales
Frequently Asked Questions
What is conversational commerce? Conversational commerce lets a customer ask a question, get a recommendation and pay inside a chat, whether that’s WhatsApp, Messenger, Instagram or an AI assistant, without leaving the conversation.
Is checkout inside Messenger available to every business? Not yet. Meta announced on October 6, 2026 that it’s testing direct checkout in Meta Business Agent on Messenger with Shopify, and similar tests in Meta AI. These are tests, so availability varies.
What is agentic commerce? It’s shopping done by an AI agent on a customer’s behalf, searching, comparing and sometimes buying. Walmart’s Agent Gateway and Agent Profiles, announced on October 21, 2026, are an example of a retailer building for it.
Is it safe to let an AI agent shop for me? It depends on the controls. Safer designs give the agent its own profile, a spending limit and approval for each purchase, and avoid sharing the customer’s own login. US banks have recently restricted outside agents that used customers’ credentials.
How should small businesses start? Begin with one chat channel, a complete product catalogue and a clear payment method, and keep a person available for complaints and larger orders.
Be Ready When the Customer Says Hello
Not every shopper will want an assistant buying for them, and chat checkout is still being tested, so none of this will happen overnight. But the direction is consistent: more selling is moving into conversations, and more shoppers, human or automated, expect clear answers without a detour.
For businesses, this means rethinking how product information is presented, how quickly customer questions are answered and how easily a purchase can be completed. Product details, pricing, availability, delivery information and return policies all need to be accurate, accessible and easy to understand. The experience must also feel trustworthy, with clear boundaries around what AI can handle and when human support is needed.
The businesses that do well will be the ones that prepare for this shift without losing sight of the customer experience. The goal is not simply to add more AI, but to make buying easier, more relevant and more reliable. If you’d like to think through what this means for your own store, our case studies show how we approach digital experiences and growth with other brands.
