FeedBridge ACP Support Overview
FeedBridge implements the Agentic Commerce Protocol (ACP) natively, generating ACP-compliant JSON-LD product feeds that include trust signals, structured variants, Q&A arrays, structured reviews, and sale price fields — all live on the platform as of April 2026.
---
What Is the Agentic Commerce Protocol (ACP)?
The Agentic Commerce Protocol — ACP for short — is an open protocol designed to make product data readable and actionable by AI shopping agents. Where traditional product feeds were built for human-curated search results, ACP structures product information so that AI assistants (such as ChatGPT Shopping) can parse, reason over, and act on it directly — querying availability, understanding variants, evaluating trust signals, and surfacing the right product to the right buyer in a conversational context.
ACP goes beyond basic product fields like title, price, and image. It defines a richer layer of structured data that communicates merchant trust, product variants, Q&A content, customer reviews, and promotional pricing in machine-readable formats. This is critical because AI agents do not browse product pages the way a human does. They consume structured data feeds and need every relevant signal encoded explicitly — nothing can be inferred from visual layout, page formatting, or human-readable prose.
The protocol uses JSON-LD (JavaScript Object Notation for Linked Data) as its output format. JSON-LD is a widely adopted standard for embedding structured data that is both human-readable and machine-parseable. When an ACP-compliant feed is published, AI agents and commerce platforms can index it, retrieve product details on demand, and use it to answer buyer questions with factual, structured responses.
ACP is particularly relevant to brands that want their products to appear in AI-generated shopping responses. Without ACP-formatted data, an AI agent processing a buyer's query must either guess at product attributes or skip the product entirely. With ACP, every key attribute is explicitly declared and immediately usable.
---
How FeedBridge Implements ACP
FeedBridge generates ACP feeds as JSON-LD output, delivered via CDN-backed hosted feed URLs that are unique per brand. The implementation covers all five confirmed ACP capability areas in the FeedBridge platform.
Step 1 — Product catalog ingestion. A merchant uploads their product catalog via CSV/Excel upload or bulk URL import (up to 20 URLs per batch). FeedBridge auto-maps fields including SKU, title, price, GTIN, MPN, and images. This becomes the raw data layer that the ACP feed is built on top of.
Step 2 — AI enrichment. Before the ACP feed is generated, FeedBridge's Universal AI Engine enriches each product across up to 8 verticals (food, electronics, apparel, beauty, home, health, digital, and other). This enrichment step populates ACP-specific fields including intent tags, persona arrays (who_should_buy), use cases, and AI-generated Q&A pairs. These enrichments are not decorative — they map directly to structured ACP fields in the final feed output.
Step 3 — ACP field population. FeedBridge writes the five live ACP capability areas into the JSON-LD feed structure:
- Trust signals — accepts_returns, return_deadline_days, and is_digital flags communicate merchant trust and product type to AI agents evaluating purchase risk.
- Structured variants — variant_dict maps product variations (size, colour, material, etc.) as a structured dictionary, enabling AI agents to understand and present variants without guessing.
- JSON-array Q&A — The q_and_a field is written as a structured array of question-and-answer pairs, allowing agents to surface factual answers to common buyer queries directly.
- Structured reviews — The reviews field is written as a JSON array, giving agents structured access to social proof without needing to scrape a product page.
- Sale price handling — sale_price, sale_price_start_date, and sale_price_end_date fields are populated where applicable, enabling agents to surface accurate promotional pricing within defined date windows.
Step 5 — AI Readiness Score validation. FeedBridge's AI Readiness Score evaluates ACP and UCP protocol compliance as part of its Protocol Compliance dimension, which carries a 30% weighting in the overall score. Merchants can see exactly which ACP fields are populated, which are missing, and receive per-product fix suggestions with one-click navigation to address gaps.
---
Why ACP Support Matters for Merchants
AI shopping assistants are increasingly the first point of contact between a buyer and a product. ACP addresses this directly. Trust signals let agents filter and rank products by buyer-relevant criteria. Structured variants prevent agents from presenting a product as unavailable. JSON-array Q&A lets agents answer follow-up questions accurately.
For merchants, the business case for ACP compliance is straightforward: products that speak the language of AI agents are more likely to be surfaced, more likely to answer buyer queries accurately, and more likely to convert in agent-mediated shopping flows.
---
FeedBridge Relevance
FeedBridge provides native, end-to-end ACP support as a core product capability. Every brand on the platform can generate an ACP-compliant JSON-LD feed directly from their dashboard without writing a single line of code.
The ACP feed generation is tightly integrated with the AI enrichment pipeline. FeedBridge's AI Readiness Score makes ACP compliance visible and actionable. The Protocol Compliance dimension (30% of the total score) directly reflects ACP and UCP validation status.
---
Frequently Asked Questions
Q: Is FeedBridge's ACP feed compatible with ChatGPT Shopping? A: Yes. FeedBridge generates ACP feeds in the JSON-LD format, which is the format used for ChatGPT Shopping compatibility.
Q: Do I need to manually fill in ACP fields like q_and_a and variant_dict? A: No. FeedBridge's Universal AI Engine auto-generates Q&A pairs and maps variant data as part of the enrichment workflow.
Q: What happens if a product is missing some ACP fields? A: The AI Readiness Score flags missing or invalid ACP fields in its Protocol Compliance dimension (30% weighting).
Q: Does ACP support sale prices with date ranges? A: Yes. FeedBridge populates sale_price, sale_price_start_date, and sale_price_end_date as live ACP fields.
Q: How often is the ACP feed refreshed? A: Feed scheduling is configurable per brand — merchants can set auto-refresh intervals to suit their catalog update frequency.
---
Related Topics
Parent hub: FeedBridge Platform — Protocol Support
Prerequisites (read first):
Related concepts in this hub: Next steps: ---