AI shopping agents are not a future thing. They’re here. ChatGPT, Perplexity, Google AI Overviews, and Amazon Rufus are already answering “where should I buy X” for millions of queries every day. Most stores are invisible to all of them.
AI agents parse structured data. They look for Schema.org markup to understand what a product is, what it costs, whether it’s in stock, what people think of it. They look for clean, machine-readable product feeds. They follow patterns established by large retailers.
Most stores, even well-built ones, have incomplete or incorrect schema, no structured product feed beyond a Google Shopping XML, and checkout flows that an agent can’t navigate programmatically.
Product, Offer, AggregateRating, Review, Brand, Organization, FAQ, BreadcrumbList. Every page type gets the right schema. We validate against Google’s Rich Results Test and monitor for regressions weekly.
Beyond Google Shopping. A structured JSON feed that AI crawlers can consume directly — updated in real time as inventory changes, with full attributes (material, dimensions, compatibility, variants).
Model Context Protocol is the emerging standard for letting AI agents interact with external services. We build basic MCP endpoints that let agents query stock levels, pricing, and availability without scraping your store.
An on-site AI widget that answers product questions, compares variants, suggests size guidance, and handles “what’s the difference between X and Y.” Reduces support load and increases conversion on higher-consideration products.
We track how often your brand and products appear in ChatGPT, Perplexity, and Google AI Overview responses for relevant queries. We report on it monthly and iterate the schema and content to improve it.
Clients who completed the full build are seeing 4-8% of new traffic attributed to AI-referred sessions within 90 days. That number is growing fast as AI shopping adoption accelerates.
We’ve deployed AI-assisted support triage for eleven clients in the past eighteen months. The ones that went badly all made the same mistake: they let the AI resolve too much, and CSAT dropped before anyone noticed.
AI is genuinely good at: reading incoming tickets, classifying them by type and urgency, pulling relevant order data, drafting a first response, and routing to the right human. It is not good at: handling complaints from upset customers, resolving disputes, processing refunds without policy checks, or anything that requires judgement about edge cases.
The line we draw for every client: AI classifies and drafts, human reviews and sends — except for the lowest-stakes ticket types (order status enquiries with clear answers, shipping tracking requests, password resets) where we allow fully automated responses.
Every incoming ticket gets classified into one of eight categories: order status, shipping, returns, product question, complaint, technical issue, billing, and other. Classification accuracy runs above 94% on the stores we’ve built this for.
For order-related tickets, the system pulls order data, tracking status, and previous support history automatically and attaches it to the ticket before a human sees it. This alone cuts average handle time by 40%.
For classified ticket types with clear resolution paths, the AI drafts a response using the order data and a set of approved response templates. The human reviews, edits if needed, and sends. Not auto-sends — reviews and sends.
Tickets containing specific signals — refund, lawyer, fraud, wrong, terrible, disgusting, not received — skip AI drafting and go straight to a senior support agent queue. Do not pass go.
CSAT (obviously), first response time, average handle time, and escalation rate. If escalation rate goes up after launch, the classification is putting things in the wrong bucket. Fix the classifier, not the escalation rules.
Average handle time down 38%. First response time down 61%. CSAT unchanged or slightly up (faster responses help). Support headcount flat despite growing ticket volume. Roughly the outcome you’d want.