Sixty-two percent of consumer packaged goods brands are actively exploring agentic shopping tools, according to Spins’ 2026 Executive Pulse survey. But 31% of those surveyed report their products are not optimized for AI agent discovery, Food Navigator USA reported.
The gap between adoption interest and data readiness exposes a structural problem: AI agents do not interpret brand positioning the way human shoppers do.
Agents Build Their Own Brand Perception
Jessica Wright, VP of Foundry at Spins (an agentic discovery solution for CPG brands), explained that AI agents assemble their own “mental model” of a brand by blending product datasets, marketing copy, reviews, and user-generated content. Brands have limited control over how agents interpret that data, she told Food Navigator.
The operational requirement is specific: agents need structured product attributes (ingredients, caffeine content, certifications, functional benefits) rather than aspirational copy. Wright illustrated the problem with a hypothetical energy drink marketed as delivering “an active sense of being throughout the day.” An AI agent parsing that copy finds no mention that the product is an energy drink, no caffeine content, no health claims. The data is functionally invisible to the agent.
“Fluffy marketing content is not going to be as easily interpretable for an AI assistant,” Wright said.
From SEO to Generative Engine Optimization
The shift extends beyond product pages. Spins data shows 9 in 10 retailers are adopting or piloting AI, and 68% of retailers look to generative AI to improve marketing and content, per Food Navigator. ChatGPT activity has skyrocketed 30 times since January 2025, according to Spate data cited in the report.
Wright described the evolution from SEO to generative engine optimization (GEO) as additive rather than replacement: optimizing for AI agent queries also strengthens traditional search performance. But brands must now translate their data into formats machines can parse, not just formats humans find persuasive.
The Data Infrastructure Layer
The competitive implication is that product data infrastructure, historically treated as a back-office function, becomes a P&L line item. Brands that invested in clean, structured attribute databases gain a discovery advantage in agentic commerce. Those relying on narrative-driven marketing without machine-readable product data risk being filtered out before a human ever sees the recommendation.
Wright predicted that utilitarian purchases like staple replenishment may eventually be fully automated by agents, making structured data even more critical for differentiation as AI assistants learn individual consumer preferences over time.