Guide

AI shopping agents. Help them find you.

Which AI shopping agents exist in September 2026, how each one reads your product data, where they fail on variants, and what to fix this week.

Try it for $1Book a call

Get your regulated media plan

Thank you! Your submission has been received!
We'll evaluate your application and shortly get back to you with the confirmation!
Oops! Something went wrong while submitting the form.
QuickAds Editorial Team
 · Reviewed by
Nitin Mahajan, Founder & CEO
 · Last updated
September 21, 2026

The short answer

An AI shopping agent is software that searches, compares and sometimes buys products on a person's behalf. As of September 2026 the live examples include Meta's Muse, the invite-only Instinct, ChatGPT's shopping surface and Google's AI Mode. Each reads merchant product data differently, and each fails differently on complex variants.

Key takeaways

  • Four agent surfaces matter to US merchants right now: Meta's Muse, launched 8 September 2026, the invite-only Instinct, which pays through Stripe's Link wallet, ChatGPT's shopping surface, and Google's AI Mode.
  • Agents get product data three ways: crawling your page in a browser, reading an intermediate product index built ahead of time, or routing fetches through a consumer device to avoid server IP blocking.
  • Complex variants are the most common failure: size plus pack count plus style together is where agents return the right product with the wrong SKU and the wrong price.
  • Some large retailers block agent traffic deliberately, because an agent turns a branded storefront into one row in a comparison table. That is an opening for merchants who stay readable.
  • Classic Google ranking does not predict agent visibility: Productrise measured 1.28% product overlap between classic Google results and AI Mode across 2M+ listings in August 2026.
  • The first fix is server-rendered product data, not content. If your price and variant matrix only appear after JavaScript runs, a crawling agent may never see them.

Make it easy for a shopper to understand why your product fits their request. That starts with clear product facts and continues through comparison, delivery expectations and the buying experience.

Test a few real purchase questions against your own catalog. Record what appears, where the facts are wrong and whether the product link works. Use the failures you can reproduce to set the priorities.

The direct answer

Every channel has a lane. None is the lane a normal DTC brand gets.

Match the format to what each platform will clear, in the right order, and paid works. Skip the order and the account gets flagged before the first learning phase finishes.

Meta

Narrow, real, certification-gated

Topical CBD runs with LegitScript certification and written approval. GLP-1 telehealth runs when it manages weight instead of curing it. Both need clean audience naming.

TikTok

Organic-first, paid is the exception

Paid CBD is closed to self-serve. Paid GLP-1 needs a TikTok sales relationship most brands do not have. Organic at scale and Spark Ads carry the channel.

Programmatic

Widest lane, least glamorous

CTV, native, audio and retail media run on publisher relationships and contextual targeting. This is usually where the real volume ends up.

Channel status matrix

Read this before a single ad gets built.

Status changes by product form (topical vs ingestible, branded drug vs telehealth program). Policy moves fast; confirm current terms before launch.

Product / claim typeMetaTikTok (paid)Google / YouTubeProgrammatic / CTVRetail media
Topical CBD (balm, salve, cosmetic)RestrictedRestrictedRestrictedAllowedAllowed
Ingestible CBD (oil, gummy, capsule)Not availableNot availableNot availableRestrictedRestricted
GLP-1 telehealth program (compounded, non-branded)RestrictedNot availableRestrictedAllowedRestricted
Branded GLP-1 pharma (named drug, on-label)Not availableNot availableNot availableRestrictedNot available
Dietary supplements (structure/function claims)AllowedRestrictedAllowedAllowedAllowed

Allowed = normal ad review. Restricted = certification, age gate or a narrower claim set. Not available = no standard self-serve path.

Straight talk

Three beliefs to drop before you brief anything.

The myth

If Meta approved my page, my ads will run fine too.

What is real

Page approval and ad approval are separate reviews. CBD and GLP-1 copy is read by policy teams looking for health claims and personal-attribute targeting, whatever your organic page has posted for months.

The myth

TikTok is where our audience is, so budget goes there first.

What is real

Your audience being on TikTok and TikTok letting you pay to reach them are two different facts. Build organic and creators there first. Paid follows the channels that accept the ad.

The myth

Programmatic is a last resort for brands banned everywhere else.

What is real

For CBD and weight management, programmatic is frequently where the real spend lives. It is less visible in a Meta Ad Library search, which is why most guides underrate it.

The sequencing

Not a menu. An order.

Skipping steps is how brands end up with a banned Meta account and no fallback channel already built.

Step 01

Lock claims and certification

LegitScript for CBD, a treat-vs-cure claims library for GLP-1, structure/function lines for supplements. This unlocks every paid channel after it.

Step 02

Seed owned and organic

Email, SMS, site content and TikTok organic build proof and the creator bench at zero platform ad-policy risk.

Step 03

Scale demand on programmatic

CTV, native and audio carry the volume once certification is in place. Usually the largest media line.

Step 04

Open the narrow Meta lane

Certified, claims-checked creative for real incremental reach, with the naming and targeting discipline the policy demands.

Step 05

Amplify TikTok with Spark Ads

Boost what already proved itself organically rather than briefing cold paid concepts the platform is unlikely to approve.

Step 06

Close the loop on retail media

Meet the demand you created at the point of purchase.

Go deeper

The category guides.

Each guide takes one category through platform rules, claims, creators and the creative lane.

CBDCBD marketing SearchCBD SEO SupplementsSupplement marketing GLP-1Weight loss & GLP-1 ClinicsWeight loss clinics LocalMed spa marketing TelehealthTelehealth marketing CreatorsInfluencer compliance
Where QuickAds fits

Creative and campaign planning for sensitive categories, inside the lane.

Use an approved claims library to brief creative, produce the variants, work with creators, and carry the learnings into the next batch.

32M+

Research-library ads behind creative intelligence and briefing.

100+

Managed creatives per month across static, video and creator formats.

See how QuickAds helpsSchedule a call
FAQ

Questions teams ask before launch.

Can CBD or GLP-1 brands still advertise on Meta?

Yes, inside a narrow lane. Topical CBD and weight-management programs can have eligible routes when certification, claims, imagery and landing pages meet current platform requirements.

Should regulated brands move their budget to TikTok instead?

Not as the default paid channel. Organic and creator content can be more useful first, with paid amplification only where the category and content are eligible.

Why does programmatic matter?

Programmatic opens display, native, CTV and audio inventory across publishers and exchanges, making it an important scaling lane for categories with tighter social-platform rules.

What is the right launch order?

Lock claims and certification, seed owned and organic, scale programmatic, open the permitted Meta lane, amplify proven TikTok content where eligible, then close the loop on retail media.

Do the same rules apply to creator content?

Yes. Creator claims still need disclosure, substantiation and a compliant brief, and boosted content must also fit the platform category rules.

Smart creatives and campaign planning for sensitive categories, from $2,000 a month.CBD · Supplements · GLP-1 · Med spa · Telehealth
See how QuickAds helpsSchedule a call

The measured gap

Classic search and AI Mode are different shelves.

Productrise compared 2M+ listings across 100k+ SERPs in the US and UK, 9–31 August 2026, running the same queries on both surfaces on the same day.

Products ranking in both
1.28%
Matched products with a different main seller
49.6%
AI Mode price premium on matched products
21.6%

Median, classic search

$100

Median, AI Mode

$149

Source: Productrise, August 2026. Google surfaces, not a forecast for any individual catalog.

Three surfaces

Ranking in ChatGPT means three different jobs.

Each is fed by a different system. Winning one does nothing for the other two, and they need different owners.

01

Written answers

Fed by live retrieval plus training memory. Won with crawler access, quotable pages and third-party coverage.

Content and SEO
02

Product results

Fed by your merchant feed and the Agentic Commerce Protocol. Won on feed quality and checkout integration.

Ecommerce and ops
03

Sponsored placement

Bought in OpenAI's Ads Manager, or through Amazon DSP. Budget and targeting rather than content.

Paid media

Two standards

Same four asks, different front doors.

Strip out the vocabulary and both protocols want the same things from a merchant. That overlap is why sequencing beats choosing.

ACP

OpenAI and Stripe · reaches ChatGPT

UCP

Google · reaches Google surfaces

  • A catalog a machine can parse, one row per sellable variant
  • Price and availability that are true right now
  • A programmatic way to confirm a cart, priced by you
  • An order lifecycle you can report back on

You stay merchant of record under both. The shared work takes a quarter; the protocol-specific work takes weeks.

The division of labour

Shopify hands you the structure. You fill it.

Ships with the platform
  • Structured product record
  • A real variant model
  • Standard product taxonomy
  • Category attributes
  • Product JSON-LD in default themes
  • Server-rendered Liquid templates
Still on the merchant
  • Category, often unset or too shallow
  • Variant option names that map
  • Barcode and identifiers
  • Metafield definitions
  • Literal description style
  • robots.txt.liquid decisions

The contract

What a machine-readable video carries.

VideoObject served in the page HTML rather than injected by JavaScript. These are the properties worth getting right.

namedescriptioncontentUrlembedUrlthumbnailUrluploadDatedurationtranscriptcaptionhasPartpublisher

Plus the association: point VideoObject.about at the product entity, or Product.subjectOf at the video, with matching @id values.

How agentic shopping works

From shopping request to shortlist

An illustrative journey. Not every shopping agent follows these steps, and not all support autonomous payment.

“Rain jacket for a Seattle trip next Thursday, under two hundred dollars, medium, nothing neon.”

At a glance

Classic SEO and LLM SEO are different games

Classic SEOLLM SEO
Unit of competitionA URL ranked for a keywordA passage quoted inside an answer
Query shapeTwo to four wordsA full sentence carrying budget, constraints and context
Result setTen results the reader chooses betweenOne answer naming three to five sources
VolatilityPositions hold for weeksThe same prompt can return different brands within the hour
Winning assetThe pageThe paragraph that still makes sense cut out of the page
MeasurementA rank trackerA fixed prompt panel you rerun and score

In this guide

What an AI shopping agent actually is
The agents that exist right now, in September 2026
The three ways agents get your product data
What breaks: complex variants and deliberate blocking
The diagnostic: find out how agents see you today
Easy fixes you can ship in under a week
The harder work, and why it pays
Where QuickAds fits

What an AI shopping agent actually is

An AI shopping agent is software that takes a shopping instruction in plain language, finds candidate products, compares them, and in some flows completes the purchase. The shopper may never open your site. The agent does, or it reads a copy of your data somewhere else.

That distinction matters more than anything else on this page. Some agents visit your product page the way a browser would. Others never touch your server and read a cached feed instead. The fix differs depending on which one is in front of you.

An agent is not the same thing as an AI search answer. An answer names products. An agent narrows a set, holds a cart, and can pay. The data requirements overlap. The failure modes do not.

At a glance: what separates an agent from a chat answer

  • It acts. It selects a variant, adds to cart and in some flows pays, rather than describing options.
  • It needs live values. A stale price kills a checkout, where it only mildly embarrasses a chat answer.
  • It picks one. Most agent flows return a single recommendation or a shortlist of three, not ten links.
  • It carries a wallet. Payment can sit with the agent's platform rather than in your checkout session.
  • It is blockable. Agent traffic often arrives from identifiable infrastructure, and some retailers already turn it away.

The practical read: treat an agent as a very literal buyer who cannot ask you a follow-up question. If the answer is not in your structured data, the agent guesses or drops you. Guessing is worse than dropping, because a guess produces a wrong price that a shopper only discovers at checkout.

This sits inside the wider shift toward agentic commerce, where the buyer on the other side of your product page is increasingly a program acting for a person.

The agents that exist right now, in September 2026

Four shopping agent surfaces matter for US merchants today: Meta's Muse, Instinct, ChatGPT's shopping surface and Google's AI Mode. The differences between them are practical, not philosophical, and they change what you fix.

Meta Muse

Meta launched Muse, a personal AI agent, on 8 September 2026. It runs in a private cloud virtual machine with a visible browser, so the user can watch it work, and it sits behind a separate permissions system called Sentinel. Pricing tiers are free, $20 and $100 per month.

For a merchant, the visible browser is the signal that matters. Muse fetches pages the way a person's browser would, which means your rendered product page, not only your feed, is what gets read.

Instinct

Instinct is an invite-only personal agent that pays through Stripe's Link wallet. The payment path is the part worth planning for: checkout can complete inside the wallet rather than through your normal funnel, so your analytics will not see a session shaped like the ones you are used to.

ChatGPT

ChatGPT's shopping surface recommends products inside the answer. Since 9 February 2026 it also carries ads: OpenAI sells sponsored product cards inside ChatGPT itself, bought through its own Ads Manager at ads.openai.com. Since 10 September 2026 there is a second door into the same surface, an Amazon Ads pilot letting select US advertisers extend campaigns into ChatGPT as conversational ads bought through Amazon DSP. Organic product mentions and paid placements now share a screen, which is new and worth watching.

Google AI Mode

AI Mode is the surface with the most merchant-visible research behind it, and the findings are uncomfortable. A Productrise study of 2M+ listings across 100k+ SERPs in the US and UK during August 2026 found 1.28% product overlap between classic Google results and AI Mode, and the main seller differed on 49.6% of the products that did match.

Ranking on page one of classic Google therefore tells you close to nothing about whether AI Mode shows you. Treat them as two channels that happen to share a company.

The three ways agents get your product data

Agents obtain product data in three ways: direct browser crawling, an intermediate product database, or fetches routed through a consumer device. Knowing which one is reading you tells you which fix will work.

1. Browser crawling

The agent loads your product page in a real or headless browser and reads what renders. This is the most current source and the most fragile one. Client-side rendered prices, variant selectors that need a click before they expose options, and cookie interstitials all cost you here.

2. Intermediate product databases

The agent queries a product index built ahead of time from merchant feeds, marketplace listings or a partner integration. Speed is the reason it exists: an index answers in milliseconds where a crawl takes seconds. Freshness is the price. A price change can take hours or days to surface.

When an agent quotes an old price, the index is usually the cause rather than the agent's reasoning. That looks like a bug to a shopper and like bait-and-switch to a regulator, so it is worth catching early.

3. Routing through consumer devices

Some agents route fetches through a consumer's own device or residential connection instead of a data center. The motive is blunt: server IP ranges are easy for a retailer to identify and block, and a shopper's home connection is not. Which products do this is not something vendors document, so treat it as a traffic pattern to look for in your own logs rather than a list you can look up.

The consequence for you is that bot filtering will not cleanly separate agent traffic from human traffic. Sessions that look human but never scroll, never hover and then check out through an unfamiliar wallet are the tell.

How to tell which one is reading you

  • Crawling: server logs show a page fetch seconds before an agent-sourced quote or order.
  • Index: the agent quotes a price you changed more than a day ago, and a fresh crawl shows nothing.
  • Device routing: residential traffic with no scroll depth, no pointer movement and a wallet checkout.

Feed quality carries the first two cases. If you have not audited yours recently, start with product feed optimization before you touch anything else.

What breaks: complex variants and deliberate blocking

Two failure modes dominate right now. Agents are still weak on complex variants, and some large retailers block agent traffic on purpose. Neither is close to solved, and both have a merchant-side response.

Complex variants

Agents struggle when a product varies on more than one axis at once. Size plus pack count plus style together is the combination that reliably breaks them. A 12-pack of medium in charcoal is three decisions the agent has to hold at the same time while reading a page built for a human who clicks.

The symptom is specific: the agent recommends the right product and the wrong SKU. It returns the single-unit price on a multipack listing, or it takes the default variant because the others only appear after a selector interaction. The shopper sees $14 in the agent and $42 at checkout, and leaves.

Shopify's guidance on agentic-ready product data names variant grouping as a first-order requirement rather than a refinement, alongside precise taxonomy, literal descriptions and real-time price and inventory. Shopify also reports AI-referred orders growing nearly 13x year over year, and AI-referred visitors converting at nearly 50% higher rates than organic search.

Deliberate blocking

Some large retailers actively block agent traffic. The logic is defensive and rational. An agent that compares five retailers turns a branded storefront into one row in a table, and a retailer with its own app would rather own the front door.

For a smaller merchant this reads as opportunity rather than threat. When large retailers shut agents out, the agent still has to return something, and what it returns is whatever stayed readable.

The risk also runs the other way. Blanket bot rules written for scrapers years ago will block agents you want. Check your WAF and your robots.txt before you conclude you have a visibility problem.

The diagnostic: find out how agents see you today

You can establish where you stand in an afternoon, with no tooling and no budget. Run these seven steps against your ten highest-margin products and write the answers down, because you will want the baseline later.

  1. Ask several agents the same buying question. Use ChatGPT, Google AI Mode, and Muse if you have access. Phrase it the way a shopper would: "best waterproof hiking boots for wide feet under $180". Record whether you appear at all.
  2. Ask for your product by name. If an agent cannot find a product when you tell it exactly what the product is, you have a data problem, not a ranking problem. This is the single most useful test on the list.
  3. Check the price the agent quotes against your live price. A gap wider than a day of staleness points to an intermediate index rather than a crawl.
  4. Ask for a specific variant combination. Request a size, a pack count and a color together, then check whether the SKU and price returned are the ones you would have picked.
  5. Load your own product page with JavaScript disabled. Whatever disappears is what a crawling agent may never see. Price and the variant matrix are the two to watch.
  6. Search your server logs for agent user agents, and for single page fetches immediately before orders. Sessions with one fetch and no scroll are your agent traffic, whether or not they self-identify.
  7. Read your WAF rules and robots.txt. Look for blanket denials by user agent string or by data-center ASN that would catch agents along with scrapers.

How to score it. Failing steps 2 or 4 means fix the data first, because presence work on a broken catalog just gets you found and then misquoted. Passing 2 and 4 while failing 1 means the data is fine and the visibility work starts. Failing step 7 means you are blocking yourself, which is the cheapest problem on this page to solve.

Repeat the same prompts monthly. Single readings are noisy, because agent answers vary by session and account, so the trend is the signal. Setting that up properly is covered in AI visibility tracking.

Easy fixes you can ship in under a week

Four changes do most of the work, and a competent developer can ship all four inside a week. Each one below has what to change and what good looks like when it is done.

Render price, stock and the variant matrix in the HTML

What to change: serve price, availability and every variant combination in server-rendered HTML rather than only after a client-side fetch. What good looks like: your product page with JavaScript disabled still shows every size, pack count and color with its own price and stock state.

Rewrite the first 200 characters of every title and description literally

What to change: replace mood copy at the top of the field with what the thing literally is, including material, quantity and fit. What good looks like: "Merino wool crew socks, 3-pack, men's US 9-12" sits where "Built for the trail ahead" used to.

Give each variant its own URL and its own structured data

What to change: expose every purchasable variant at a stable URL carrying its own price, GTIN and availability, grouped under the parent product. What good looks like: you can link an agent straight to the 12-pack in medium charcoal and it resolves to that SKU with the right price.

Fill the taxonomy and attribute fields you have been leaving blank

What to change: populate product category, material, size system, age group and pack quantity as real fields rather than facts buried in prose. What good looks like: no required attribute in your feed spec is empty across your top 100 SKUs.

Unblock the agents you want

What to change: replace blanket bot denials with an allow list for the agents you have decided to serve, and log the rest rather than dropping them silently. What good looks like: you can name which agents you accept and show the log line proving they got through.

None of these need new content. They are data and delivery changes, which is why they are fast.

The harder work, and why it pays

Three things take longer than a week. They are also the reason some merchants will be readable to agents in a year and others will still be arguing about it.

Fix the variant model, not the variant page

Most variant failures come from a data model that treats a multipack as a separate product, or that encodes pack count inside the title string. Restructuring so that size, pack count and style are three independent attributes on one parent is a migration, not an edit, and it usually touches your ERP as well as your storefront.

Budget weeks, not days. The payoff is that every downstream surface, feeds, agents, marketplaces and your own search, stops guessing at the same problem separately.

Get price and inventory to real time

Nightly feed exports were acceptable when a human saw the price on your page before buying. An agent may quote whatever it last cached. Moving to event-driven updates on price and stock removes the most damaging class of agent error, which is a confidently stated wrong price.

Make your media parsable

Agents cannot watch a video. If your best product explanation lives in a 30 second clip with no transcript, no structured metadata and no schema, that explanation does not exist as far as an agent is concerned. Tagged catalog video for AI search exists to close that gap.

Build the measurement habit before you need it

You cannot tell whether any of this worked without a baseline taken before you changed anything. The diagnostic above is the baseline. Run it, store it, and repeat it on a fixed cadence.

Be realistic about the timeline. This channel has real failure rates today, and the honest reason to do the work now is that the fixes are the same ones that make your feed, your marketplace listings and your on-site search better regardless of what agents do next.

Where QuickAds fits

QuickAds works on three parts of this problem. It is worth being plain about which parts, and which we do not touch.

Shopify catalog fixes

We do variant grouping, taxonomy cleanup, rewriting product fields into literal machine-readable language, and live price and stock sync. This is the work that decides whether an agent returns your correct SKU or your default one, which is the failure described earlier on this page.

Catalog video with correct tagging

We generate product video per SKU with structured metadata and schema attached, so agents and AI search surfaces can parse what is actually in the video instead of seeing an opaque file.

Performance creative

100+ creatives per month, 5-7 day turnaround, with creative intelligence trained on 32M+ ads. That matters here because agent surfaces now carry paid placements next to organic recommendations, and both need assets that hold up. Details are on creative as a service.

The wider point is coverage. We run creative intelligence, strategy, production and campaign management as one chain. Analytics-only tools tell you what happened and stop there. Premium production shops make assets and never touch your product data. Agent readiness fails in the seams between those two, which is why we work across them.

What we will not tell you: that agent traffic will lift your revenue by some number. Nobody can honestly forecast that yet, and we cannot make a retailer-style block on someone else's platform disappear. What we can commit to is throughput on the work itself, at software from $199 per month or managed from $5,000 a month or 5% of ad spend, whichever is higher.

Quick readiness check

Can an agent actually read your catalog?

Eight checks drawn from this guide. Tick what is already true for your catalog.

0 of 8Tick the checks that apply

Frequently asked questions

What is an AI shopping agent?

An AI shopping agent is software that takes a shopping request in plain language, finds and compares candidate products, and in some cases completes the purchase for the user. Unlike a chat answer that only names options, an agent selects a specific variant, holds a cart and can pay, which is why accurate live price and variant data matter so much.

Which AI shopping agents can actually buy things right now?

As of September 2026, Meta's Muse runs a visible browser in a private cloud virtual machine and launched on 8 September 2026 with free, $20 and $100 tiers. Instinct is invite-only and settles payment through Stripe's Link wallet. ChatGPT and Google AI Mode recommend products inside answers, with checkout behavior varying by merchant integration.

Do AI shopping agents crawl my site or use my product feed?

Both happen, and a third path exists. Some agents crawl your product page in a browser and read what renders. Some query an intermediate product index built ahead of time from feeds, which makes them fast but stale. Some route fetches through consumer devices to avoid server IP blocking. Your server logs will tell you which.

Why do AI shopping agents pick the wrong product variant?

Because most product pages only reveal variants after a click, and most catalogs encode pack count inside the title instead of as a field. When size, pack count and style vary together, the agent holds three decisions at once and often falls back to the default variant. The result is a right product with a wrong SKU and a wrong price.

Should I block AI shopping agents from my website?

Some large retailers do, because an agent turns a branded storefront into one row in a comparison table. For most merchants that logic does not apply, and the bigger risk is blocking agents by accident with old scraper rules. Check your WAF and robots.txt, then allow the agents you want and log the rest.

Make these in QuickAds

Try it for $1