How does a seven-day sprint make your online shop agent-ready?
An agent-ready shop states what each product is, what it costs, whether it is in stock and how it ships as data an AI assistant reads without guessing. The Machine Readability Sprint adds that layer in seven days on top of the platform you already run, tests it with real assistants and leaves your checkout untouched.
What can an agent do with your shop after the sprint?
After the sprint, an agent can get a list of your pages and their full text in one request, read each product's price, stock, shipping and return terms as data, open your content pages without banners or scripts, and ask your shop directly for search results and stock. Every piece arrives as code or CDN settings, tested before handover.
| Deliverable | Where it lives | What it changes |
|---|---|---|
llms.txt and llms-full.txt, an index of your site for AI tools, see llms.txt for ecommerce | Site root | An agent finds all your pages and reads their full text in one request |
JSON-LD @graph, your product facts as structured data | Every product, category and policy page | An agent reads product, offer, seller, shipping and returns as fields |
| Markdown mirrors, plain-text copies of your pages | Same path as each content page, with .md | An agent reads the page text without layout, scripts or banners |
| HTTP Link headers, pointers to those copies | CDN or server configuration | An agent finds the plain-text copy from the page response itself |
| MCP tools, the standard agents use to query a shop, see the protocols compared | Hosted endpoint, reading your product API | An agent searches your catalogue, checks stock and reads offers |
| UCP or ACP routing, the protocols agents use to check out | Your checkout, where your channels use them | An agent hands a ready order to your checkout, the first step of agentic checkout with Convert |
| Validation report | Handed over on day seven | You see every test we ran, with the assistant and the result |
How does the sprint work without replatforming?
Your shop keeps running as it is: the checkout, the product database and the theme stay untouched, and removing the layer returns everything to its previous state. On most platforms the layer is a theme snippet or app plus a small worker at your CDN that serves the plain-text copies and headers.
- Shopify and Shopify Plus
- A theme snippet for the product data, an app proxy for the plain-text copies and llms.txt
- WooCommerce
- A plugin that turns your product data into JSON-LD and serves the plain-text copies
- Adobe Commerce (Magento)
- A module for the product data, with the plain-text copies served from the CDN
- Salesforce Commerce Cloud
- A cartridge for the product data, an edge worker for the plain-text copies and headers
- SAP Commerce (Hybris)
- Edge worker in front of the storefront, fed by the product export
- Headless and custom stacks
- Edge worker or middleware, reading your product API
Which product data does an agent-ready shop declare?
An agent compares offers on a handful of fields, and each one has a Schema.org property that states it without guessing. For a Swiss shop that means the price in CHF with VAT included, the delivery time to a Swiss address and the return terms, declared on the page a person sees, so data and text agree.
| What the agent checks | Schema.org property | Swiss detail to declare |
|---|---|---|
| Price | Offer.price, priceCurrency, PriceSpecification.valueAddedTaxIncluded | CHF, with VAT included, as on the page |
| Stock | Offer.availability on each variant | Stock for the size or colour asked for, not for the product as a whole |
| Identity | gtin, brand, mpn | The GTIN from the manufacturer or your GS1 Switzerland number |
| Variants | ProductGroup with hasVariant and variesBy | One declared variant per size and colour, see sizes and variants for agents |
| Delivery | shippingDetails with deliveryTime and shippingDestination | Delivery days to an address in Switzerland, and the carrier you ship with |
| Returns | hasMerchantReturnPolicy with merchantReturnDays and returnFees | Return window and who pays for the return |
| Seller | Organization with name, address and contact | Your company and its Swiss address |
What access is needed to start the sprint?
Your team gives us read access and a kickoff session, and we do the rest: the build, validation, deployment and tests, with a progress report at the end of each day. That means read-only access to your product feed or catalogue API, permission to add DNS or CDN rules and one session with your technical owner.
- Read-only access to the product feed or catalogue API
- Permission to add rules at your CDN or DNS provider
- A staging environment, or approval to deploy behind a feature flag
- One kickoff session with your technical owner
- A named person to approve the final deployment
What do teams ask before booking the sprint?
Developers and e-commerce leads usually ask four things before reserving a Sprint: what happens during the seven days, which platforms it runs on, what their own team has to do and how it relates to Discover. The answers below let the kickoff start with the access already in place.
What happens during the seven days?
We turn your catalogue into structured product data, publish llms.txt and plain-text copies of your pages, and set up the agent tools listed in the deliverables table, then check everything against the readiness tests before handover.
Which platforms does the sprint support?
Shopify, WooCommerce, Adobe Commerce, Salesforce Commerce Cloud, SAP Commerce and custom stacks, as listed in the compatibility section, all without moving your shop to a new platform.
How much does it cost to make an online shop agent-ready?
The Sprint is a fixed fee per shop, quoted after a first conversation, because the work depends on the platform, the number of products and variants, and the languages and markets you sell in. A Buyer Journey Check beforehand tells you which stage loses the most sales, so the seven days go to the fix that pays back first.
Do we need a product feed for ChatGPT?
It depends on your platform. OpenAI states that Shopify merchants are included in ChatGPT shopping through Shopify Catalog without extra work, and that other merchants can apply to send a product feed directly. The Sprint makes sure the feed, the structured data on the page and the visible page carry the same price, stock and delivery terms, since a mismatch is a reason for an agent to drop the offer.
Should we let GPTBot and other AI crawlers into the shop?
For product and policy pages, usually yes, since an assistant cannot name or compare what its crawler cannot read. The Sprint reviews your robots.txt crawler by crawler, separating search and shopping crawlers from training crawlers, so you decide each one on purpose. Our guide to llms.txt, Markdown and schema for online shops explains what each crawler reads.
What does our team need to do?
Grant the access listed above and name one developer who can review and merge changes. The rest of the work runs on our side.
How does the sprint relate to Discover?
The Sprint builds the readable layer once. Discover is the monthly retainer that keeps it current as the catalogue changes and measures whether assistants name you more often.
Book the week your shop becomes readable
If you already know where agents drop you, the sprint fixes it in a week. If you do not, the Buyer Journey Check tells you which parts of the sprint to run first.