Connect once. The system runs every week.
The first run establishes a baseline for your catalog, engine answers, and purchase path. SoldByAI then runs the same real shopping questions through the leading AI engines every week and turns errors and gaps into proposed fixes. You approve each change before it moves forward, and the next cycle checks the result.
- 1One-time connection
- Connect a product source
- Approve the catalog draft
- 2Initial baseline
- Product catalog
- Engine answers
- Purchase path
- 3Weekly cycle
- Test
- Proposed fixes
- Your approval
- Next-cycle retest
Start with your existing product sources
Fastest: one URL.
SoldByAI builds a machine-readable catalog draft from your existing product sources. The draft brings product identifiers, specifications, price, availability, and source details into one consistent structure for your approval.
Storefront URL
Read public product information and create a catalog draft
Platform connection
One-click authorization for Shopify and WooCommerce
Feed URL
Connect an existing feed for daily sync
Excel or CSV
Import after field mapping
- Product source
- Catalog draft
- Your review and approval ✓
- Machine-readable catalog
The approved catalog feeds four supported outputs: SoldByAI subdomain product pages, the Google Merchant Center (GMC) feed, the Agentic Commerce Protocol (ACP) feed, and the agent catalog.
- SoldByAI subdomain product page
- Google Merchant Center (GMC) feed
- Agentic Commerce Protocol (ACP) feed
- Agent catalog
Your storefront, URLs, checkout, and design stay unchanged.
Your catalog, your call.
The first run establishes the baseline
Your catalog, engine answers, and purchase path are measured separately. Original answers, test times, and failure points remain available for every later comparison.
- Catalog baseline
- Product identifiers, structured data, price, and availability are checked for completeness.
- Engine baseline
- The leading AI engines are tested one by one. The China market's major engines are on the same sweep. Each result is recorded as a mention, citation, recommendation, or absence.
- Purchase-path baseline
- Four gates, recorded one by one: find, compare, select, then verify the purchase route is live.
- Coverage
- Tested at
- Original answer
- Evidence level
- Failure point
Every cycle builds on the last
Each week, SoldByAI runs the same real shopping questions across the leading AI engines and compares every result with the existing baseline. New evidence and fix outcomes carry into the next cycle.
Your optimization agent gets to know your brand with every cycle.
It carries forward your brand facts, competitive context, proven shopping questions, and completed fixes. Each new cycle avoids work already done and focuses on what remains unresolved.
Every proposed fix includes the original evidence, an action path, and a retest status.
Brand-specific knowledge
- Brand facts
- Competitive context
- Sweep history
- Proven shopping questions
- Fix history
- TestThe same real shopping questions
- CompareCurrent results against the baseline
- ProposeProduct · original evidence · action path
- ApproveEvery change requires your approval
- RetestCheck the result in the next measurement cycle
Testing continues after a recommendation
A recommendation is only the start. Each cycle runs an agent through find, compare, and select, then verifies the purchase route is live. Link reachability is tested separately, and each failure point is tied to a proposed fix.
Each order, attributed to its engine. Orders placed through the SoldByAI endpoints carry the buyer query with them.
Product links from each of four supported outputs carry source parameters into your storefront. Your existing analytics can identify the output and product behind each visit.
- Source output
- Product
- Source parameters
- Merchant analytics
- Find
- Compare
- Select
- Route live
- Checks passed
- Failure point
- Link reachability
- Tested at
SoldByAI runs the cycle. The decisions stay with you.
Your decisions
- Select the product source and connection method
- Approve the catalog draft
- Decide which fixes go live
- Apply changes that belong in your storefront
SoldByAI's weekly cycle
- Build and check the machine-readable catalog
- Run sweeps across the leading AI engines
- Keep original answers and prepare proposed fixes
- Check the result of each fix and rerun the purchase path in later cycles
Every change requires your approval.
Check your catalog for machine-readable product data
Enter your store URL. The audit checks product identifiers, structured data, and other key signals.