More than 900,000 Walmart associates use a conversational AI assistant every week, and between them they ask it over 3 million questions a day. AI in retail has moved well past the press-release stage, and the biggest names are now using it to answer product questions, support associates, help customers shop and check purchases.

That's good news if you're choosing tools for your own team. These deployments give you a real benchmark: the job each tool handles, how widely it's available and what the retailer has reported about its use. Just remember that a shopping assistant and an automated receipt check solve very different problems, even when both get filed under "AI."

This guide walks through live deployments in stores and online, a couple of pilots and one project Amazon has scaled back, so you can see what each one teaches before you commit your team's time and budget.

Key takeaways:

• Large retailers already run AI at scale: Lowe's across more than 1,700 stores, Home Depot in all 2,000-plus U.S. stores and Walmart with more than 900,000 weekly associate users.
• Rollout size shows reach. Repeat usage and a measured outcome show whether a tool fits the working day.
• Tools that suggest and systems that act need different approval rules, so check permissions feature by feature.
• Start with one recurring task, run a limited trial and judge it against the goal you set before launch.

First, let's pin down what "live" actually means.

What counts as a live deployment?

A live deployment is an AI system in active use in a retailer's operations or customer experience. We count what each retailer says is running today, and a planned rollout stays in the "planned" column until it's complete.

Each type of evidence answers its own question.

If you've sat through a vendor demo packed with impressive numbers, this is how you sort them.

Keep this table handy as you read the examples below. It turns every product demo into a sharper conversation.

Which retail AI tools are live?

Retailers are deploying AI for three broad jobs: associate support, shopping guidance and purchase verification. Here's how each looks in practice, with the reported figures linked to their sources.

Lowe's helps associates answer product questions

Lowe's uses Mylow Companion to help associates find product details, inventory information and project advice on their sales-floor devices. The assistant launched across more than 1,700 stores on May 5, 2025, according to the Lowe's rollout announcement.

Picture an associate covering an unfamiliar department on a busy Sunday.

They can ask a question in everyday language, including by voice, and keep the conversation with the customer going while the tool fills in the details.

For your team, the starting point is the information gap. Which questions regularly send an associate hunting for a colleague or digging through product documentation?

Home Depot brings guidance into stores

Home Depot's Magic Apron helps shoppers locate products and get guidance tailored to their chosen store. Shoppers reach it through Store Mode in the app or a QR code on store signage.

The assistant is live across all 2,000-plus U.S. stores, and Home Depot reports that Magic Apron handles millions of questions monthly in its in-store expansion announcement.

This one is worth studying if your customers struggle to find products or understand compatibility. Before adding something similar, check whether your product and location data can answer those questions reliably (connecting that data is the heart of a unified commerce platform).

Walmart reports regular associate usage

Walmart reports heavy use of its conversational assistant among associates. Its June 24, 2025 update recorded more than 900,000 weekly users and over 3 million daily queries in the associate AI announcement.

For your own rollout, measure repeat use alongside access.

Often it's something small, like one login too many.

Sam's Club verifies purchases at the exit

Sam's Club uses computer vision at the exit to verify purchases, which reduces the need for manual receipt checks. Walmart's April 2025 investor materials confirmed the system had reached the whole club fleet in less than a year, in its investment community presentation.

This system automates a defined part of the store visit and changes the staff process around it. When you evaluate something comparable, include how your team handles exceptions.

So ask the supplier to show you what happens when a purchase can't be verified. Your associates will need that answer long after the demo ends.

Zara offers virtual try-on online

Zara's virtual try-on generates images of a shopper's avatar wearing real products. Inditex reported availability in 43 markets and more than 7 million sessions in its March 2026 results statement.

If you're considering virtual try-on, decide which problem you're testing first.

Helping someone picture an outfit calls for a different evaluation from helping them choose the right size.

Amazon's assistant can complete purchases

For eligible products, Amazon's Alexa for Shopping overview describes Buy for Me completing purchases on a customer's behalf using the shopper's primary address and credit card.

That raises a bigger approval question than a tool that only suggests products. When you assess purchasing features, ask exactly what the customer authorizes and how they review or cancel an action.

Check permissions feature by feature, too. Adding something to a cart and completing payment carry very different risks.

Some projects are still finding their feet, though, and those deserve a closer look.

Which projects need closer attention?

A project can teach you plenty while it's still running in only a few places.

Sephora described its ChatGPT app as a U.S. pilot in its March 2026 announcement. Payments and checkout were listed as future updates, so the announcement covers product discovery inside the app while purchasing is still to come.

Target confirmed that teams were using Store Companion in daily work in its October 2025 technology update. The update describes daily use without a store count, which makes it a good prompt for your own supplier conversations: how many stores use the tool today?

These descriptions reflect what the cited sources say at the time of publishing, and products move fast (sometimes faster than their press pages).

For a buying decision, ask for current availability and for references from retailers using the exact features you're considering.

What has Amazon scaled back?

Amazon announced plans to close its Amazon Go and Amazon Fresh physical stores while continuing to offer Just Walk Out technology elsewhere. Its store strategy update reports the checkout-free technology operating in more than 360 third-party locations across five countries.

That split matters when you're weighing a similar investment, especially if someone has just forwarded you the headline with a worried note. Judge a store format's economics on their own, then look at how well the technology would fit your setting.

So ask how a supplier's reference sites compare with yours.

Your stores deserve their own business case.

What should your team test first?

Test a task that causes a recurring problem and has a result you can observe. The examples above are good prompts, but your priorities should come from your customers and store teams.

Your associates already know where the day gets stuck. Ask them.

1. Choose the task. Ask associates where they lose time or leave customer requests unfinished, then pick one problem to solve.
2. Record the current process. Note how the task gets done today and what a successful result looks like before you introduce AI.
3. Set the approval rules. Identify which suggestions a person reviews and which actions the system may take on its own.
4. Run a limited trial. Include ordinary shifts and difficult cases, then review results and staff feedback before expanding.

Say post-visit follow-up keeps slipping. You could test whether associates can use customer notes to prepare well-timed messages more consistently, then check the quality of the messages as well as how many go out.

Give associates an easy way to flag a wrong answer or an unusable draft.

A tool that creates a second queue of work is the opposite of what a busy store team needs.

Endear's guide to planning a retail AI pilot can help you turn that task into a defined trial.

How should you assess the results?

Assess results against the goal you set, with a comparison that shows what changed. Treat higher usage as an early signal and judge the tool on better service or more profitable sales.

It's tempting to celebrate the first chart that goes up and to the right. Give it a second look first.

Suppose AI users buy more than non-users. Were they already your most likely buyers? Promotions, staffing changes and the customer mix can all nudge the outcome too.

Ask who collected the figures, what period they cover and which costs are included. Endear's retail AI vendor questions give you a solid starting list for that conversation.

Where does Endear fit?

Endear applies AI to the part of retail where relationships turn into repeat sales: customer outreach. Its AI Opportunity Engine ranks every customer in your database by opportunity and delivers a daily queue, so associates start each shift knowing who to reach out to and why. The reasons come from signals such as a recent purchase, an abandoned cart, a lapsed regular or an upcoming birthday.

Those signals are only as complete as the customer data behind them. That's why many groups start by connecting their POS systems to one customer record.

For each opportunity, the AI drafts an on-brand message using your brand voice guidelines and templates. The associate reviews it, edits the wording if they like and decides whether to send it by email or SMS. Nothing goes out to customers without a human approving it.

One specialty retailer went from 60 outreach messages a week to 360 within six weeks of turning on the AI Opportunity Engine, with no added headcount, while conversion rates held steady.

In a Censuswide survey of 1,000 U.S. consumers commissioned by Endear, 55% said they have made a purchase because of follow-up communication, as reported by CX Dive.

The Opportunity Engine gives you a clear workflow to test in a pilot: relevance of the suggested customers, time spent reviewing drafts and how customers respond. Agree on who owns customer-facing messages before your trial starts, and you'll have clean results to judge.

Frequently asked questions

Retail AI tools vary in how much they automate and how teams measure their value, and that's where most questions start.

How is AI used in retail stores?

AI supports product questions, store navigation, purchase verification and customer outreach. The right application depends on the task you need to improve and the information the system can access.

Does a rollout prove business value?

A rollout confirms availability at the reported scope. To judge value, pair it with usage and outcome measures for your own team.

Can AI work without human approval?

Some systems perform defined actions on their own, while others prepare suggestions for a person to review. Check permissions at the feature level, including how errors and cancellations are handled.

Which retail AI tool should you choose?

Match the tool to a clear operational need, your available data and your team's working process. Then test it with the people who'll use it before committing to a wider rollout. If you're comparing clienteling platforms across many stores, our guide to enterprise clienteling software lists the evidence to request.

How should you measure a pilot?

Measure the outcome you chose before the trial and track adoption alongside it. Include staff feedback and the time spent correcting errors, so your assessment captures the work the tool creates as well as the work it saves.

Pick one task and give it an owner

The retailers above started with a single job, whether that was answering product questions or checking receipts, and scaled from there. You can do the same: choose the retail problem you want AI to help solve, name someone to run the trial and give them clear criteria to continue, change or stop.

If customer follow-up is the job, see how Endear's AI Opportunity Engine would fit your team's day. Book a demo and walk through a morning queue, from the ranked opportunities to the approved message.