AI Feed Agent

AI Agent for Product Feed Management: Setup, Audits, Troubleshooting and Optimisation

Examples:

Loading conversation…

AI Feed Agent•now

Help centre, with steps you can apply
You can ask questions and receive answers based on feed attributes and channel requirements, e.g., title, product categories, product variants, shipping, pricing, and more.
Attribute mapping suggestions
You can use the AI agent to assist in suggesting a way to approach a feed attribute mapping based on "we have this in the source" and what you would want to happen in the case of a lack of information.
Rule drafts you can validate
You can input what you want to happen with your feed, e.g., "append size to titles for apparel variants," and receive a draft of what you could use for a rule, what you might want to consider, and a way to validate before you roll it out.

Find missing and weak attributes
AI Feed Agent identifies areas of missing or poor information such as brand, GTIN, product type, color, and material. The issues are grouped together for easier identification of what to fix first.
Spot inconsistencies and anomalies
You can use the AI Feed Agent to detect conflicting values in product variants, unusual changes in price, unusual settings in shipping, and incorrect category assignments compared to similar products.
Check against channel requirements
The AI Feed Agent also points out common areas of disapprovals and reduced visibility such as truncated titles, restricted terms, poor product type, missing information, and incorrect attribute formats.

Feed Processing

Data Enrichment

Channel Sync

AI Optimisation

Rule Engine

Title Rewrites

Run source imports
Start an import for a selected source and verify it is complete. Review what changed, including new items, updates, and items that fell out before they flow downstream.
Run feed syncs
Start a feed sync and send updates to channels with the latest information. Monitor status and catch errors before they affect exports.
Block items
Prevent selected products from being exported if needed, such as policy issues, stock status, or price issues. Record blocked items and reasons for blocking them.
Unblock items
Allow blocked products to be exported after issues are resolved. The next import or sync will pick them up and allow them to flow into relevant feeds again.
Schedule updates
Schedule source imports and feed syncs to occur on your chosen schedule with history of past runs, upcoming runs, and errors.

Root cause analysis that you can follow
Understand exactly what's changed in your feed and its impact with attribute diffs, unexpected value changes, and trends in your products, categories, brands, or variants.
Explain errors in plain language
Turn feed errors into actionable reasons with clear next steps, so non-technical teams can take action without wading through diagnostic screens.
Prioritise what to fix first
Prioritise feed errors by severity and reach, for example, what prevents approvals, what negatively impacts matching results, or what causes inconsistent product listings across platforms.
Cross-check in Google Merchant Center
Verify item status, diagnostics, and policy or quality signals in the Merchant Center to see exactly what's happening from Google's perspective and whether it's a feed problem, crawling problem, or account problem.
Validate landing page structured data
Fetch product pages to validate key structured data like Product and Offer microdata, prices, availability, GTINs, and compare it with your feed to quickly identify discrepancies before they cause errors or disapprovals.

Setting up your Feedoptimise account…
Preparing your account and workspace settings.

Explain the current setup
Answer "what does this do?" in a very practical way, including what sources go into the feed, what each transformation does, and what rules apply to what products.
Rule-by-rule handover notes
Create basic documentation on each rule, including what it does, what it is meant to achieve, and what to do with it.
Guided onboarding with guardrails
Guide new team members through common activities with safe defaults, including what to do first, what to change, and when to ask for a review.
Learn by doing, safely
Use examples to teach team members what to expect as outputs, with quick tests to verify that changes are correct before they affect the catalogue as a whole.

Field-level validation
Format, length, required fields, and other constraint-related issues are identified and highlighted for you. Risks of failure and potential impact on coverage and relevance are identified and proposed solutions are offered for you to implement.
Variant consistency checks
Validate that variants differ where they should, for example size or colour, and match where they must, such as brand and identifier rules. Flag cases where variants may collapse into duplicates, with suggestions to resolve.
Evidence checks for AI-enriched content
Checks are carried out for consistency of enriched information and patterns, and risks of unsupported information are highlighted for you to decide what is safe and what is not.

★ ★

APPROVED

Date: ____/____/____
____________________
SIGNATURE

In stock items

5.3k

Good titles

99%

Good descriptions

98%

Items on sale

33%

Identify optimisation opportunities
Highlight products with low relevance signals, few attributes, or unclear categorisation, and then recommend which fields to optimise first based on potential impact.
Segment the catalogue in a useful way
Organise products by intent and similarity as well as category, to enable more targeted rules for example by using different title formats for accessories compared to main items.
Explain what changed and why
Make feed changes more actionable by linking to results achieved by showing changes to attributes, affected products, and emerging trends.
Build a practical roadmap
Turn findings into a plan of action with quick wins, structural changes, and reusable templates that can be implemented consistently as the catalogue grows.