Bain Says Agentic Commerce Is the Next Retail Revolution — And Retailers Must Choose: Embrace AI Agents, Restrict Them, or Build Their Own

Bain & Company has published what may be the most important strategic framework for AI commerce to date. In "Agentic AI Commerce: The Next Retail Revolution Is Here," the consulting firm draws a direct parallel to the 1990s — when retailers who resisted e-commerce fell behind and those who embraced it survived. "When e-commerce first took off in the 1990s, many established retailers faced an existential crisis: resist or embrace? Those that prioritized brick-and-mortar identities only over an online presence eventually fell behind." The same choice, Bain argues, now faces every retailer in 2026. AI agents "poised to significantly transform how consumers discover, research, compare, and purchase products" are moving from demo to deployment. The ultimate form — full agent-to-agent commerce, where the buyer's AI and the retailer's AI transact directly — "could shortcut or bypass classic retail websites altogether."

The three strategies

Bain lays out three responses on a spectrum.

Embrace the agents. Allow third-party AI agents — ChatGPT, Gemini, Perplexity, and the next generation of agentic storefronts — to crawl your site, list your products, and close transactions through their interfaces. Bain argues this "may be the smarter play for retailers that do not have the brand awareness or consideration in their target audience to drive traffic at scale." The trade-off is commoditization: your products become interchangeable with competitors inside an AI's comparison answer.

Build your own agents. Bain cites Home Depot's Magic Apron as the proprietary model. The advantage is access to "customer and purchase data that third-party agents lack," enabling "more personalized service that taps into existing consumer trust." Amazon's Rufus and Alexa+ follow the same playbook: keep the entire discovery-to-transaction journey inside the retailer's walls. The cost is AI engineering talent most retailers do not have on staff.

Restrict and negotiate. Control which agents get access to your catalog and negotiate data-sharing agreements with the ones you allow. Walmart has taken this path — building Sparky as a proprietary AI while blocking external agent-initiated checkouts. Shopify is in a similar posture, restricting external agents while launching its own Agentic Storefronts.

Bain's conclusion is direct: "The winning strategy will likely not be a binary choice of staying open or closed to third-party AI agents, but rather a blend of both — or even building their own proprietary agents."

The data control question

The strategic pivot point, according to Bain, is data. "Retailers can keep control of customer data and loyalty only if they control where and how customers complete their transactions." If a consumer discovers a product through ChatGPT and completes the purchase through ChatGPT, the retailer loses the search query, the browsing behavior, the loyalty enrollment, and the customer service relationship. If the consumer discovers through ChatGPT but clicks through to the retailer's site to buy, the retailer keeps all of it. This is exactly the split OpenAI pivoted to after Instant Checkout failed in March: discovery inside ChatGPT, transaction on the retailer's own checkout page. Amazon's approach is more aggressive — refuse external AI commerce entirely and build Rufus to handle both ends of the journey internally.

The fraud complication

Bain flags a problem most coverage has missed. "Many current e-commerce fraud prevention technologies are built to detect and block bots." AI agents, by their nature, behave like bots — automated, high-volume, pattern-based. Retailers now have to distinguish malicious bots from legitimate AI shopping agents acting on a consumer's behalf. New authentication methods and tokenization frameworks will be needed. DataDome has identified the same challenge in travel, where AI agents are hitting airline and hotel APIs in ways existing fraud systems treat as attacks.

Mubboo's Take

Bain's three-strategy framework perfectly maps what we have observed this week. Amazon chose Strategy 2 — build Rufus and Alexa+. Walmart is blending Strategy 2 (Sparky) with Strategy 3 (restrict external agents). OpenAI's failed Instant Checkout was an attempt to make Strategy 1 the default, and it collapsed because retailers were not willing to surrender transaction control. For Mubboo, we operate alongside all three strategies without threatening any of them. We do not crawl retailer sites. We do not compete with proprietary retail AI. We produce independent editorial content that any AI agent — third-party or proprietary — can cite when answering consumer questions. Our value sits upstream of the strategy choice on mubboo.com/shopping and mubboo.au/shopping: we inform the decision, regardless of which checkout path the consumer ultimately takes.