Getting started

Expanding Web prepares Adobe Commerce and Magento stores for AI shopping agents: the assistants inside ChatGPT, Perplexity, Google and the merchant-side protocols they speak (UCP, ACP, MCP). We work from Poland, mostly with US merchants, and everything below is bought and delivered without a sales call.

1. Free agent-readiness scan

Enter your storefront URL on the home page. The scanner runs the eight public checks live against your store and shows the result on the page in under a minute. If you leave an email address, an engineer reviews the result and sends a short report with the three risks worth fixing first — within two business days. Nothing is installed on your store and nothing is charged.

What each check measures: Readiness scanner checks.

2. Agentic Commerce Readiness Sprint — $4,500, five business days

The sprint is bought directly on the page through Stripe Checkout; billing details and tax identifiers are collected at payment. It is one price, no discounts, and there are two slots per month — the page shows whether the current month still has one.

What you get:

• catalog and product-data audit against what agents actually read,
• a target architecture for ACP, UCP and MCP on your store,
• one working prototype on your data,
• a security and observability plan,
• a production roadmap with effort estimates,
• if you already run an AI chatbot or shopping assistant: a review of it (grounding, price and stock accuracy, escalation paths) at no extra cost.

What happens after payment, day by day: After you buy the sprint.

3. Production implementation

Quoted after the sprint, from the roadmap it produced — typically $10,000–$20,000 depending on catalog size and how much of the stack we build versus configure. Never sold before a sprint, because the estimate would be a guess.

4. Feed & Protocol Watch

From $399 a month, priced by catalog size and level of service: ongoing technical maintenance of what AI agents read and call on your store. Every deliverable ends in a technical recommendation or a fix:

• Feed health — product data and merchant feeds checked against the current OpenAI (ACP) product feed, Google Merchant Center and UCP catalog requirements; you get a mapping fix list for what fails.
• Endpoint checks — scheduled re-runs of the UCP profile and MCP initialize checks, with an alert when an endpoint stops answering or stops conforming, plus the reproduced cause and a recommended fix.
• Monthly protocol-watch note — changes to ACP, UCP, MCP and merchant-feed specifications that affect your store, and what we change in response.
• Quarterly re-scan — all eight agent-readiness checks re-run against the previous baseline, with a prioritised fix list.

An AI citation check on buyer-intent prompts is included as context, not as the product. This is not an AI-visibility dashboard: tools such as Profound, Peec AI and the Semrush AI Toolkit measure what AI says about you; we keep your store transactable for AI agents. No uptime SLA, and no guarantee of citations, rankings or acceptance into a platform programme.

Who you deal with

Every engagement has a named technical account manager — Michał Kucharek. He answers within four business hours on business days (Monday to Friday, Central European Time), sends a written status at every phase, and checks in 30 days after handover. Marcin Dancewicz, the founder, signs the architecture and is your escalation on the same terms.

Contact: info@expandingweb.com, (+1) 415 802 1072 (answered around the clock), (+48) 693 868 369 (European hours). Reference implementation you can try an agent against: ready.expandingweb.com.