D-Congress 2026 produced one clear signal from the HCL Commerce team: AI is dominating the commerce roadmap, and B2B is behind.

My take is the same. And I want to show where the gap actually sits for HCL Commerce+ stores - because the answer is not in the backend.

The HCL recap of D-Congress was honest about the picture: Agentic AI is moving fast. The customer-facing use cases (LLM-based shopping assistants, conversational search, ChatGPT Commerce integration, AI-driven product recommendation) are live in consumer commerce and arriving in B2B. The question is how existing HCL Commerce+ deployments get from "our backend handles AI-enriched data" to "our storefront is actually agent-ready."

That transition is not a backend question. HCL Commerce+ already has backend AI features - Discover+ for behavioral data, Marketing Cloud integration for personalization signals. The infrastructure for feeding data to agents exists. What is missing is the layer those agents operate on: the frontend.

What Agentic Commerce Needs from the Frontend

An agent-ready storefront is not about adding a chatbot to a JSP page. The architectural requirements go deeper:

Structured data that agents can parse. ChatGPT Commerce, Google AI Overviews, and Perplexity Commerce do not crawl for text - they parse structured Schema.org markup. Product availability, pricing, specifications, and delivery timeframes need to be encoded in machine-readable format at the page level. An Aurora-Storefront with server-side rendering can implement Schema.org, but it cannot do it at the speed the GEO Management Agent needs - updating markup dynamically as inventory changes, as contract pricing updates, as promotions go live.

Clean APIs that agents can call. AI shopping assistants (the kind that will run inside ChatGPT or as browser extensions) need to call into your catalog, your pricing, your availability in real time. A composable frontend that exposes clean, cacheable API endpoints for product data is the integration surface. Aurora-Storefront's rendering model does not expose this surface cleanly.

Component-level control for dynamic layouts. When a performance agent detects that a specific product category page has LCP regression, it needs to be able to adjust the component configuration - swap a heavy image carousel for a text-first layout, change the hero component breakpoint, adjust the lazy-loading threshold. That requires a component-model frontend, not a monolithic template.

Fast iteration for agent-driven tests. A/B tests driven by AI agents need deployment cycles measured in minutes, not sprints. If the agent identifies that a checkout-flow variant converts 12 percent better, the winning variant needs to go live before the insight goes stale. Studio-based component switching enables that. Sprint-based frontend changes do not.

The D-Congress Observation, Extended

The HCL team observed at D-Congress that B2B commerce is behind consumer in AI adoption. That observation is correct. Part of the reason is structural: B2B commerce has prioritized workflow correctness (pricing, contracts, order management) over experience velocity. The backend is the strength; the frontend was the thing that got built to satisfy the backend.

That is exactly why the agentic bridge for HCL Commerce+ is a frontend-layer problem. The backend can expose structured data - HCL Commerce+'s REST APIs (/wcs/resources/store/...) already return pricing, catalog, and inventory in structured format. What you need on top is a frontend that can consume and expose that data in the format agents expect, iterate at agent speed, and maintain brand consistency while agents run tests.

Laioutr's Agentic Frontend Management Platform is the layer that operationalizes the D-Congress signal. The five Frontend Agents - Content, SEO/GEO, Performance, Vertriebssteuerung, and the underlying Larry AI - run on top of the same component library that marketing teams use in Studio. HCL Commerce+ feeds the data. Laioutr's agents optimize how that data is presented and found.

The Decoupling Prerequisite

Before any of the agentic layer is possible, the decoupling has to happen. An Aurora-Storefront that is tightly coupled to HCL Commerce+'s JSP rendering cannot run a GEO agent. It cannot run a performance agent at component level. It cannot expose the clean API surface that AI shopping assistants need.

The prerequisite for agentic readiness is frontend decoupling: separating the frontend layer from the backend so each can evolve at its own speed. HCL Commerce+ continues to handle everything it handles today. Laioutr's Frontend Management Platform provides the storefront layer that supports agentic tooling.

USP 1 (Decoupling) is not just about today's time-to-market. It is about architectural optionality for what comes in the next 24 months.

What This Looks Like in Practice

For an HCL Commerce+ deployment with Laioutr as the frontend layer, the agentic setup works like this:

The GEO Management Agent monitors AI Overview visibility for product categories and updates Schema.org markup when new products are added, prices change, or inventory shifts. The SEO Management Agent watches internal linking patterns and flags when a new category page is orphaned. The Performance Monitoring Agent tracks LCP on the highest-GMV category pages and triggers a component-configuration review when LCP crosses the 2-second threshold.

None of this requires backend changes. HCL Commerce+ keeps running. The agents operate on the frontend layer.

For more on the complete frontend layer picture for HCL Commerce+, the Hub Post covers the full architecture. For the SEO/GEO product specifics, see SEO and GEO on Laioutr. For the broader industry signal, the Agentic Commerce tech stack overview from May 2026 covers the full stack picture.

The 30-Minute Question

If you are running HCL Commerce+ and want to understand what agentic readiness would look like for your specific setup - which agents apply first, what the frontend-decoupling sequence is, and what the realistic timeline is - a 30-minute discovery call is the right next step.

30-minute Discovery: How would a headless frontend for your HCL Commerce+ setup look concretely?