Agentic commerce gets operational fast

Self-improving agents and product data governance emerge as two of the most pragmatic AI developments for brands and retailers, reports Miya Knights, Retail Technology Publisher

If Retail's Big Show, held by the US National Retail Federation in New York last month, had a single through-line, it was that "agentic" is no longer shorthand for a better website chatbot.

The more credible vendor conversations focused on practical software: how to translate messy, real-world customer intent into measurable outcomes, and how to keep product truth coherent as artificial intelligence (AI) systems increasingly mediate discovery, evaluation, and purchase.

Two retail technology providers—EnviveAI and Akeneo—illustrate how quickly the narrative is hardening into architecture. The first is building a self-improving "intelligence layer," designed to orchestrate multiple cooperative agents across the marketing funnel. The latter is betting that product information management (PIM) is becoming the governance layer for machine-led commerce.

"Glorified chatbots" to outcomes

In Retail Technology magazine's NRF 2026 Innovations Showcase tour report, Aniket Deosthali, co-founder of EnviveAI, positioned the company as deploying multiple AI agents "to drive the business outcomes that our customers care about". He described automated merchandising and product discovery to improve on-site search performance and visibility within generative engines, with a claimed "two to three times" increase in conversion lift.

That framing also appears in Deosthali's broader messaging. In EnviveAI's Series A announcement, he argued that the agentic era will punish shallow implementations: "Brands need more than just wrappers around LLMs [large learning models]. They need a system that continuously learns from real-world behaviour and drives the outcomes they care about. That's what we are building: self-improving agents with performance, control, and safety at their core."

Deosthali's premise is that ecommerce teams are being pulled in two directions at once: traffic is fragmenting towards generative discovery, while internal teams are still largely structured around channel siloes and manual rules. He emphasised the "intersection of consumer behaviour and AI" and the push to "modulate based off outcomes" across funnel stages —such as awareness, browse, and consideration—rather than optimising each channel independently.

Compliance as a design constraint

EnviveAI's own positioning makes that explicit, describing an "intelligence layer" that integrates signals across touchpoints and "orchestrates cooperative, brand-aligned agents to drive outcomes across the funnel.” For most retailers and brands, the subtext is important: if agentic shopping surfaces accelerate the customer journey, the winners will be the organisations that can measure what is happening, adapt fast, and keep the experience compliant and on-brand.

EnviveAI is also leaning into brand safety as a differentiator. In its Coterie case study, the company frames safety and compliance as central to performance, noting that inaccurate or noncompliant claims can risk trust and lead to regulatory penalties in heavily regulated categories. Even where retailers are not operating under category-specific regulation, many merchant operators will recognise the practical implication: as AI becomes a "front door" to shopping, governance and guardrails are moving from legal checklists to conversion levers.

Product data becomes the execution layer

Akeneo's NRF-adjacent message was complementary: agentic commerce only works if product data is structured, governed and activatable at speed.

In its Winter Release announcement, Romain Fouache, CEO of Akeneo, stated: "The new era of commerce is defined by AI-mediated interactions that move faster than human teams can manually track." He added: "Velocity isn't about rushing with AI; it's about the institutional confidence to execute at the pace the market now demands."

Akeneo is explicitly translating that thesis into its products. The release introduces an Akeneo Digital Showroom, a Stripe Agentic Commerce Suite partnership, a native Server integration with the open-source Model Context Protocol (MCP) from OpenAI, and Custom Components—designed to reduce the lag between product enrichment and activation as autonomous agents play a larger role in the buying journey.

Fouache's "what it means" warning is stark, and will resonate with retail leaders who have spent years under-investing in product content and data governance. He said: "In this new economy, your product data is no longer just a description; it is your brand's frontline salesperson. If that data isn't structured, trusted, and instantly accessible to AI models, your brand effectively ceases to exist in the autonomous buying funnel." 

Why this matters in 2026

Taken together, these two, different narratives suggest a more grounded post-NRF reality for all retailers and brands.

First, agentic commerce is rapidly becoming an integration-and-measurement problem, not a conversational user interface problem. EnviveAI is arguing for a unified outcome layer that can learn from behaviour and tune actions in real time.

Second, PIM is being redefined as part of AI’s governance infrastructure. Akeneo is effectively claiming that the organisations with the cleanest product truth—and the fastest execution loop—will be the ones that remain "discoverable" and "buyable" when AI agents, not humans, are the primary navigation layer.

Both hint at the next standards debate that will shape vendor roadmaps: interoperability among emerging commerce protocols (e.g., Google's Universal Commerce Protocol and MCP).

Even without betting on a single standard, the direction of travel is clear: retailers will want agents that plug into existing stacks, work through application programming interfaces (APIs), and prove impact against specific outcomes—conversion, revenue per visitor, returns reduction, customer satisfaction—rather than "engagement" in abstracted agentic outcomes.