Intelligence for the Age of Agentic Commerce From Around the Web
Signal
TheoryNXT launches KASPER - measures Brand Inclusion Rate across ChatGPT, Gemini, and PerplexityWalmart: Sparky-attributed GMV up 150% in Q1; baskets 35% larger than non-Sparky usersBloomreach: 75% of consumers used AI to shop in the past six months; 42% use AI shopping tools dailyAkamai: AI bots now make up nearly 48% of global commerce trafficSalesforce: retailers running their own shopper agents grew sales 59% fasterSalesforce: AI influenced 20% of global online holiday sales, worth ~$262BBain: grocery pricing and promotions must be rebuilt for price-comparing AI agentsTheoryNXT launches KASPER - measures Brand Inclusion Rate across ChatGPT, Gemini, and PerplexityWalmart: Sparky-attributed GMV up 150% in Q1; baskets 35% larger than non-Sparky usersBloomreach: 75% of consumers used AI to shop in the past six months; 42% use AI shopping tools dailyAkamai: AI bots now make up nearly 48% of global commerce trafficSalesforce: retailers running their own shopper agents grew sales 59% fasterSalesforce: AI influenced 20% of global online holiday sales, worth ~$262BBain: grocery pricing and promotions must be rebuilt for price-comparing AI agents
Stablecoin infrastructure firm Rain launched the Agentic Payments Alliance on Aug. 18 with 26 founding members, including Visa, Mastercard, Fiserv, and Circle, to set shared standards for AI agent identity, authorization, and payments before regulators do. McKinsey projects $3–5 trillion in global agentic commerce by 2030 — the rails are being built now, and CPG brands should watch which standards win.
Walmart, Amazon, and Target all reported AI assistants driving materially bigger orders this earnings season — Sparky users at Walmart spend 40% more per order — but the gains are concentrated almost entirely in repetitive grocery and staples categories, not discovery-driven retail. Early evidence that AI assistants reward habitual repurchase categories first.
Nestlé is using AI to mine clinical research and identify nutrient combinations for GLP-1 drug users, already shipping products like Boost Advanced Nutrition Shake and Milo PRO High Protein built on the findings. A concrete example of a major CPG using AI for product development speed, not just marketing or discovery.
NielsenIQ found 51% of U.S. shoppers used AI to shop in the past month — the first time adoption has crossed the halfway mark. Days later, the FTC chair said companies can't blame their AI agents' actions on the agent itself, six major banks demanded audit trails for every autonomous purchase, and Amazon shut Meta's Muse agent out of its marketplace. Adoption and accountability arrived in the same week.
Amazon Ads launched a pilot letting select U.S. advertisers buy ChatGPT ad placements through Amazon's own demand-side platform, with Delta Vacations among the first testing it, though OpenAI's own system still decides which ads actually appear. For CPG brands, it's a new, still-narrow door into paid visibility inside AI conversations, layered on top of the organic-discovery battle already underway.
Target rolled out an AI Photo Search tool that lets shoppers find products from an uploaded image, alongside a Review Insights feature that summarizes product reviews and has already lifted conversions and cart adds. For CPG brands, it's another sign that how well your product images and reviews are structured is starting to directly shape whether AI surfaces you first.
Retail Dive · · 3 min
Subscribe
Sharp, independent intelligence on agentic commerce for CPG brand leaders. Updated frequently. Subscribers get a curated briefing every Thursday. Free.
No spam. Unsubscribe anytime.
Sponsored
By the Numbers
150%
Q1 growth in Sparky-attributed value of goods sold at Walmart
Walmart, July 2026
41%
Consumers who'd rather shop through an AI assistant than a brand's website
Bloomreach, July 2026
48%
Share of global commerce traffic that is now AI bots
Akamai, July 2026
59%
Faster sales growth for retailers running their own shopper agents
Salesforce, July 2026
About
The Invisible Shelf is a free briefing for CPG brand leaders navigating the rise of agentic commerce.
AI shopping agents are becoming the new retail shelf, recommending, filtering, and buying on behalf of consumers. This publication tracks what matters.
Updated frequently. Subscribers get a briefing every Thursday. Free forever.
TikTok introduced a conversational AI shopping assistant that remembers shopper preferences across a conversation, plus one-click checkout directly from the For You feed, integrating with Salesforce, Shopify, and Stripe. The move is aimed at keeping users from turning to ChatGPT for product questions, building on the $15.8 billion TikTok Shop generated in U.S. sales in 2025 and raising the stakes for CPG brands already juggling multiple AI-native storefronts.
A new PYMNTS analysis argues retail competition is shifting from physical shelf space to who controls clean, real-time product data feeds and APIs, with digitally influenced sales already topping 60% of retail activity. It points to Apparel Group's overhaul of its entire operating model, running autonomous forecasting, pricing, and scheduling across 85 brands and 2,500 stores, as an early look at the data infrastructure CPG suppliers will be expected to plug into.
PYMNTS mapped the fast-multiplying list of agentic commerce protocols merchants now have to track, from OpenAI and Stripe's Agentic Commerce Protocol to Google and Shopify's Universal Commerce Protocol, Visa's Trusted Agent Protocol, and Mastercard's Agent Pay. Its warning that no single standard will likely win out means CPG brands betting their integration budget on just one path risk losing visibility with shoppers on a competing AI platform.
Constructor rolled out "Agentic Checkout," a Stripe-powered feature that lets shoppers discover products, ask questions, and complete a purchase entirely inside its AI Shopping Agent without ever leaving the conversation. For CPG brands sold through Constructor's retail partners, including Sephora, it's a sign that the next competitive front in agentic commerce is closing the gap between AI-driven discovery and an actual finished sale.
A new analysis draws a sharp line between "brand agents," which feed AI systems a company's official product information, and "personal agents" like Meta's Muse or ChatGPT's Dots, which shop on a consumer's behalf, and profiles startups like DaVinci Commerce building knowledge graphs to translate catalog data into language AI can use. For CPG brands, the piece argues fluency in this emerging brand-agent layer, not just shelf presence, will decide whether AI assistants describe and recommend a product accurately at all.
New data compiled by Digiday shows AI shopping assistants recommending markedly different products depending on which retail partnerships sit behind them, turning the assistants into a new gatekeeper that "picks products by rules marketers don't yet understand." For CPG brands the stakes are concrete: traffic referred by AI agents already converts at a 53% revenue premium per visit, yet only 59% of CMOs have started optimizing for AI search, leaving most brands' visibility inside these platforms a matter of chance rather than strategy.
Salesforce unveiled expanded Agentforce capabilities at Dreamforce, including an AI shopping assistant called "Carter" that can guide a customer through product discovery, comparison, and checkout while pulling real-time inventory and customer data from digital and physical stores alike. The rollout lands alongside a survey finding only 30% of retailers have fully unified their customer data across channels, a gap CPG brands' retail partners will need to close before AI agents can reliably represent products at the point of sale.
OpenAI launched a virtual try-on tool inside ChatGPT that lets shoppers upload a photo and preview how clothing and accessories will look before buying, with items organized in a new ChatGPT Library for later reference. For CPG and fashion brands, it's a direct play on cutting returns and lifting conversion by moving product visualization into the AI chat itself, rather than leaving it to a brand's own site.
Mastercard released "A Short History of the Future of Shopping and Payments," projecting that one in 10 online shoppers will routinely delegate purchasing decisions to AI agents by 2030, worth hundreds of billions in AI-driven spending a year. For CPG brands, the report frames "agent-readiness" — verifiable claims, standardized product data, and algorithm-legible presentation — as a new competitive requirement alongside traditional loyalty and shelf strategy.
Despite forecasts that agentic commerce could reach $1 trillion in US retail by 2030, voices like Hedera AI Studio's Lindsay Walker argue today's e-commerce infrastructure can't yet support autonomous AI purchasing, pointing to account logins, payment fields, and location logic that agents still can't reliably handle. For CPG brands weighing how fast to invest, it's a useful counterweight to the hype, a reminder that identity, payments, and regulatory frameworks still need rebuilding before agents can shop at scale.
A new Spreedly survey found 31% of Gen Z shoppers now trust ChatGPT most for product research versus just 12% for Google Search, and 63% of Americans said they'd let AI pick the brand if it saved them 20%. For CPG brands, it's fresh evidence that loyalty is turning fluid under AI-mediated discovery, with algorithmic value comparisons threatening to override habitual brand preference.
Shopify expanded its WebMCP support with new get_checkout, update_checkout, and complete_checkout tools, letting any browser-based AI agent — not just a named partner — read and complete a checkout directly instead of scraping or screenshotting the page. It's a sharply different bet than Amazon's agent lockouts, positioning Shopify's merchants as structurally easier for outside AI agents to buy from as the checkout layer itself turns machine-readable.
FTC Chairman Andrew Ferguson said businesses remain fully liable for what their AI shopping and service agents do, warning that "giving software the power to act does not make the software responsible when something goes wrong." For CPG brands piloting agentic commerce tools, it's a signal that documenting the authority, limits, and instructions given to any AI agent will matter as much as the technology itself.
Nvidia launched an Open Agent Safety Platform that pairs software guardrails with independent hardware monitoring able to quarantine a misbehaving AI agent within milliseconds, with more than 100 organizations including Anthropic, Microsoft, and SAP already testing it. The trust infrastructure reaches beyond banking: PYMNTS notes 132 million U.S. adults have now made AI-assisted retail purchases, yet 59% of those still end up completing checkout at Amazon rather than a brand's own site.
Adobe forecasts AI-driven traffic to U.S. retail sites will jump 130% year over year this holiday season, with spikes of 159% on Thanksgiving and 95% on Black Friday, inside a record $275.1 billion online shopping season. Categories like personal hygiene (+150%) and household cleaning (+49%) show some of the largest AI-driven gains, a sign CPG brands should have product data and stock availability ready for AI-assisted holiday stock-up trips.
Google is piloting direct in-app checkout for Flipkart products through Gemini and AI Mode with a limited group of Indian users, with a broader rollout planned for October ahead of the country's peak shopping season. For CPG brands, it's an early sign that AI assistants may start controlling checkout itself rather than just discovery, and that a retailer's financial ties to the AI platform could shape which products get frictionless purchase paths.