AI shopping agents are autonomous software programs that browse product listings, compare options, and complete purchases on behalf of consumers without direct human input during the transaction. This matters for ecommerce sellers because when an algorithm — not a person — decides what gets bought, the visual and structural quality of a product listing becomes the single most important factor in determining whether a sale happens.
On November 6, 2026, Amazon filed a civil complaint in the U.S. District Court for the Western District of Washington accusing Perplexity AI of building an agent that secretly logs into user accounts, scrapes marketplace data, and executes purchases on Amazon without authorization. According to the Bloomberg report on the filing, the Comet browser's shopping mode bypasses Amazon's robot-detection systems and violates the platform's Conditions of Use. The complaint alleges that Comet "floods Amazon's storefront with automated traffic" designed to mimic human behavior.
What Amazon Actually Alleges
The lawsuit targets three specific behaviors of Perplexity's Comet agent, as Reuters detailed in its coverage of the case. First, the agent allegedly accesses product detail pages at a rate that exceeds any realistic human browsing pattern. Second, it captures pricing and inventory data that feeds back into Perplexity's answer engine, creating a competing shopping recommendation surface. Third, the agent completes checkout flows using stored user credentials, which Amazon characterizes as unauthorized access to a commercial computer system under the Computer Fraud and Abuse Act.
Perplexity's chief business officer Dmitry Shevelenko pushed back publicly, telling TechCrunch that Comet operates "the way a human assistant would," and that blocking AI agents would punish consumers who rely on them for accessibility and time savings. The company argues that Amazon's real complaint is competitive: Perplexity's shopping recommendations redirect traffic and commissions away from Amazon's own Rufus assistant.
"The complaint is dressed up as a robot policy case, but it is really a complaint about competition in AI-driven shopping." — Perplexity spokesperson, in a statement to The Verge on November 7, 2026.
The New Rules of AI-Mediated Commerce
Whether Amazon wins or loses in court, the lawsuit confirms something sellers should have seen coming: AI agents are no longer experimental. They are a primary shopping channel. Salesforce's State of Commerce report found that 39% of online shoppers used an AI assistant to research or buy a product in the previous twelve months, and that number climbs every quarter.
When an agent shops on behalf of a consumer, the decision tree collapses. The human is not browsing 30 tabs, comparing colors, or zooming in on stitching. The agent is reading structured data, parsing image tags, evaluating listing completeness, and ranking options. Listings that lack clean product imagery, transparent backgrounds, and clear hero shots lose to listings that have them — even when the underlying product is identical.
Why Product Imagery Is the New Battleground
AI shopping agents are visually trained. They rank images the same way image-recognition models do: clarity, contrast, isolation of the subject, absence of background noise, and consistency across a set. A cluttered lifestyle photo with props, watermarks, or busy backdrops gives the agent less usable signal than a clean, well-lit, isolated product shot. The agent cannot infer quality from a moody flat-lay the way a human shopper might.
This is where most seller catalogs fall short. A BigCommerce product photography analysis showed that 61% of marketplace listings fail basic image-quality thresholds: low resolution, inconsistent framing, cluttered backgrounds, or missing the white-background main image that Amazon, eBay, and Walmart all require for primary slots. An AI agent filters these listings out before a human ever sees them.
How Sellers Can Prepare for an Agent-First Marketplace
The good news is that closing the visual gap no longer requires a studio, a photographer, and a week of turnaround. Modern tooling can produce marketplace-ready imagery in minutes. Here is the workflow that sellers adopting AI-first pipelines are running today.
- Step 1 — Audit your catalog. Pull your top 100 SKUs by revenue and check each main image against marketplace white-background specs. Flag any with busy backgrounds, low resolution, or inconsistent aspect ratios.
- Step 2 — Strip and standardize backgrounds. Use an automatic background removal tool for clean product cutouts to isolate every hero image. Consistent white or transparent backgrounds give AI agents a clean signal to parse.
- Step 3 — Render lifestyle and contextual variants. A single product often needs three to five images: a hero shot, a scale reference, an in-use shot, a detail close-up, and a packaging shot. An AI product photography studio can produce these variants from one source image.
- Step 4 — Generate platform-specific mockups. Amazon requires 2000x2000 pixels. Instagram prefers vertical 4:5. TikTok Shop wants 1:1 with text-safe zones. A product mockup generator that exports to multiple aspect ratios handles this without reshoots.
- Step 5 — Re-submit and monitor agent traffic. Watch your server logs and marketplace analytics for unusual user-agent strings. Tools like Perplexity's Comet, OpenAI's Operator, and Anthropic's browser-use agent all identify themselves in headers.
What the Lawsuit Does Not Solve
Amazon's legal action does not slow down the broader shift to agentic shopping. OpenAI launched Operator in early 2026, Google expanded Gemini-in-Chrome shopping flows, and a wave of vertical AI agents — from Stitch Fix's style bot to Instacart's meal-planning agent — already mediate billions of dollars in annual GMV. Even if Amazon wins an injunction against Perplexity, the underlying behavior pattern (an AI agent acting on a consumer's behalf) is already the default for a meaningful slice of the market.
Sellers who treat this as a platform dispute are missing the point. The dispute is about who controls the recommendation surface. But the buying surface — your product page, your images, your structured data — is yours to control. The brands that invest in clean, agent-readable catalogs now will own the next decade of AI-mediated commerce.
Frequently Asked Questions
Why did Amazon sue Perplexity's Comet AI agent?
Amazon sued Perplexity on November 6, 2026, alleging that the Comet browser's AI shopping agent violates Amazon's Conditions of Use by scraping product pages at non-human speed, capturing pricing and inventory data, and completing purchases through stored user credentials. The complaint invokes the Computer Fraud and Abuse Act and targets what Amazon calls "flooding" of its storefront with automated traffic that mimics human behavior to evade detection systems.
What does the Amazon vs Perplexity lawsuit mean for ecommerce sellers?
For ecommerce sellers, the lawsuit is a signal that AI shopping agents are a permanent, mainstream channel — not a curiosity. Roughly 39% of online shoppers used an AI assistant to research or buy a product in the past year, according to Salesforce's State of Commerce report. Sellers who optimize their listings for AI readability (clean images, structured data, consistent backgrounds) will rank higher in agent-driven recommendations, regardless of how the lawsuit resolves.
How do AI shopping agents evaluate product listings?
AI shopping agents evaluate listings using a combination of structured data (titles, bullet points, attributes) and visual signals (image clarity, background isolation, framing consistency, absence of text overlays). They rank images similarly to image-recognition models, prioritizing clean, isolated product shots over busy lifestyle photography. Listings with cluttered backgrounds, low resolution, or watermarks are typically filtered out before a human shopper ever sees them.
Do ecommerce sellers need to change their product photography for AI agents?
Yes. AI agents read images differently than humans. A clean, well-lit product shot on a white or transparent background gives the agent a stronger signal than a moody flat-lay. Sellers should audit their top SKUs, remove backgrounds from hero images, generate contextual variants (lifestyle, scale, detail, packaging), and export platform-specific aspect ratios. Modern tools can complete this pipeline in minutes rather than the days a traditional studio shoot requires.
Get Your Catalog Ready for Agent-First Shopping
The Amazon–Perplexity lawsuit is the opening shot, not the war. AI agents will keep shopping, and your product imagery is the front line. Build a clean, isolated, marketplace-ready image set for every SKU before the next wave of agents goes live.
Prepare your catalog for the agent-first era
Strip backgrounds, render studio shots, and export platform-perfect mockups in minutes.
Try Rewarx Free