AI Feed Agent

AI Feed Agent for daily feed management, setup, audits, troubleshooting, and optimisation

Examples:

Loading conversation…

AI Feed Agent•now

Help centre, with steps you can apply
Ask questions in plain language and get practical guidance tied to feed attributes and channel requirements, such as titles, product categories, variants, shipping, and pricing.
Attribute mapping suggestions
Translate “we have this in the source” into a mapping approach for the feed, including reasonable fallbacks and what to do when data is missing or inconsistent.
Rule drafts you can validate
Describe the outcome you want, for example “append size to titles for apparel variants”, and the agent proposes a rule outline, common edge cases, and a simple way to test before rollout.
More consistent decisions across the catalogue
Standardise values and handle ambiguous fields in a repeatable way, so conventions stay consistent even as product data and inventory change.

Find missing and weak attributes
Flag empty, low-quality, or non-compliant fields, such as brand, GTIN, product_type, colour, and material. Issues are grouped so it’s easier to prioritise what to fix first.
Spot inconsistencies and anomalies
Detect duplicates, conflicting variant values, unusual price changes, unexpected shipping settings, and category assignments that don’t match similar products.
Check against channel requirements
Highlight common causes of disapprovals and reduced coverage, including truncated titles, restricted terms, weak product types, missing identifiers, and incorrectly formatted attributes.
Pull structured attributes from existing content
When source data is thin, extract usable attributes from descriptions or product page content to fill gaps, for example pattern, fit, ingredients, capacity, or compatibility.

Feed Processing

Data Enrichment

Channel Sync

AI Optimisation

Rule Engine

Title Rewrites

Run source imports
Trigger an import for a selected source, confirm completion, and review what changed, including new items, updates, and products that dropped out before data flows downstream.
Run feed syncs
Start a feed sync to push the latest data to channels, track status, and catch failures early so exports stay reliable.
Block items
Exclude specific products from exports when needed, for example policy concerns, stock status, pricing issues, or data quality. Keep a record of what was blocked and the reason.
Unblock items
Re-include products once issues are resolved. The next import or sync will pick them up so they can reappear in the relevant feeds.
Schedule updates
Set scheduled source imports and feed syncs to run at the cadence you choose, with visibility into previous runs, upcoming runs, and any failures.

Root-cause analysis you can follow
Summarise what changed and what it impacts, including attribute-level diffs, sudden value shifts, and patterns across products, categories, brands, or variants.
Explain errors in plain language
Turn technical feed errors into clear reasons and next steps, so non-technical teams can act without wading through diagnostics screens.
Prioritise what to fix first
Rank issues by severity and reach, for example what blocks approvals, what reduces matching quality, and what creates inconsistent listings across channels.
Cross-check in Google Merchant Center
Review item status, diagnostics, and policy or quality signals in Merchant Center to confirm what Google is flagging, and whether the issue is feed data, crawling, or account configuration.
Validate landing page structured data
Fetch product pages and validate key structured data, such as Product and Offer markup, price, availability, and GTIN, then compare it with the feed to spot mismatches before they trigger 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 practical way, including how sources flow into feeds, what each transformation changes, and which rules affect which products.
Rule-by-rule handover notes
Generate simple documentation per rule, including intent, inputs, before and after examples, edge cases, and guidance on when to update or disable it.
Guided onboarding with guardrails
Walk new team members through common tasks using safe defaults, including what to change first, how to preview impact, and when to ask for a review.
Reusable playbooks and checklists
Create internal runbooks for recurring work, such as imports, syncs, troubleshooting, and monitoring, so knowledge is shared and easy to find.
Learn by doing, safely
Use practical examples to explain expected outputs, with quick checks to confirm changes are correct before they affect the wider catalogue.

Field-level validation
Check formats, lengths, required fields, and common constraint issues. Flag what is likely to fail downstream and what may reduce coverage or relevance, with suggested fixes for you to apply.
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
Cross-check enriched attributes against source data and known patterns, and flag anything that looks unsupported or inconsistent. Suggest safer alternatives and consistent terminology, you decide what gets applied.

★ ★

APPROVED

Date: ____/____/____
____________________
SIGNATURE

In stock items

5.3k

Good titles

99%

Good descriptions

98%

Items on sale

33%

Identify optimisation opportunities
Flag products with weak relevance signals, thin attributes, or unclear categorisation, then suggest which fields to improve first based on expected impact.
Segment the catalogue in a useful way
Group products by intent and similarity, not only by category, so rules can be more targeted, for example using different title structures for accessories versus core items.
Explain what changed and why
Link feed edits to outcomes by showing which attributes changed, which products were affected, and what patterns appeared after the update.
Build a practical roadmap
Turn findings into a prioritised plan, including quick fixes, structural improvements, and reusable templates that can be applied consistently as the catalogue grows.