When customers ask AI what to buy, does it recommend you?
Buying journeys start in ChatGPT, Claude, Perplexity and Gemini now. We make your brand the answer: discoverable, accurate, and a clean handoff into your own checkout.
Built for a specific kind of brand
£1m to £100m in revenue
Large enough that the AI channel moves real money. Not so large that a six-figure enterprise integration is the answer.
On Stripe, or close to it
Stripe is the fastest path to ACS readiness. A stack that can expose your catalogue cleanly works too.
An ecommerce function in place
They run the store and flip the platform toggles. We work alongside them and make the store intelligible to agents.
If you're earlier than that, the audit is still worth doing. The full build probably isn't yet, and we'll say so.
The shelf moved. Most brands aren't on it.
Discovery moved
Customers ask ChatGPT, Claude, Perplexity and Gemini what to buy. The recommendation happens before they ever see your site.
Checkout stayed yours
In-chat checkout retreated in 2026. The settled model is discover in AI, buy on your own site. Your store still closes the sale, if agents can find it.
Structure is the ticket
Agentic storefronts and structured feeds are moving from edge case to baseline. Agent-readable product data is a commerce prerequisite now, whatever platform you're built on.
Stripe ACS readiness
Stripe's Agentic Commerce Suite connects AI-driven discovery to a brand's live checkout and payments. JD Sports deployed it first, going live in January. The channel is in production, not on a roadmap.
Upstream of the agentic commerce standards
The governed brand record, the MCP endpoint and the structured feed are what any agentic commerce integration reads from, not just Stripe's ACS. We prepare that integration path as part of the build, so when you switch a rail on, the data behind it is already right.
You don't need an expensive consultancy to get ready for this. It's a specialist skill, not a difficult one, and it fits inside eight weeks.
Eight weeks, three steps
Audit
Weeks 1 to 2. We measure what AI says about your category today. You get the baseline whether or not you continue.
Build
Weeks 3 to 6. Feed, Claude Skill, MCP endpoint, schema fixes. Built against your live catalogue.
Activate and verify
Weeks 7 to 8. Storefront on, endpoints live, queries re-run. The before and after report lands.
The onboarding: get on the shelf
Everything an agent needs to find you, read you, and recommend you. Fixed price, fixed scope, verified result.
Marketing context engine
One governed source of brand truth: verified claims, product logic, comparisons, policies and voice. Compiled once, served to every AI surface and protocol.
- The governed brand record: the claims, comparisons and policies agents can safely quote
- Structured ACP/UCP product feed, kept accurate to price, stock and range
Agent interfaces
The working surfaces agents call, from discovery through to your own checkout.
- A Claude Skill that puts your catalogue inside Claude, and an MCP endpoint any agent can call
- On-site schema fixes so what agents read matches what you sell
- Agentic Storefront and platform toggles configured to our spec, usually by your own ecommerce function, with us alongside
Verification and accountability
The build starts and ends with measurement, so you can hold us to the result.
- AI visibility audit: real buying queries across ChatGPT, Claude, Perplexity and Gemini, recorded as your baseline
- The verified before and after report on recommendation share and accuracy
The retainer: stay on the shelf
Catalogues change. Standards shift. Assistants update. The retainer keeps your AI presence accurate without you thinking about it.
Monthly monitoring report
The same buying queries, re-run every month. You see your recommendation share move.
Feed and Skill updates
New products, price changes and range edits flow through to the feed, Skill and endpoint.
Remediation and protocol watch
When an assistant gets you wrong, we fix it. When the standards shift, and they will, we move you with them.
The Center Parcs test
This is one test, run on one brand, inside its own booking experience. We asked Claude to book a Center Parcs short break, the same request a real customer would make. Claude couldn't complete it: only booking aggregators had agent access, and Center Parcs itself was unreachable inside its own booking journey.
The visibility audit runs the same test on your category, inside your own booking or checkout experience. You see exactly what AI does when a customer tries to buy from you, verified, before we build anything.
- Claude was asked to book a real Center Parcs break.
- It answered from the aggregators it could read, not the operator.
- Center Parcs was unreachable inside its own booking experience.
- The booking would have routed through someone else entirely.
The same audit runs on our own brand, against the feed, schema and MCP endpoint this site publishes at /for-agents. The baseline comes first. Results published when we have them.
Straight answers
Will customers be able to buy inside ChatGPT or Claude?
Not today, and we will not promise it. In-chat checkout retreated in 2026 and the settled model is discover in AI, buy on your own site. But agentic commerce is on the way — the rails are being laid now in ACP/UCP feeds, MCP endpoints and Agentic Storefronts, and this work puts your brand on them early. What we deliver today is honest: discoverable and accurate in AI answers, with a clean agent-assisted handoff into your own checkout — and already standing on the rails when agents start transacting.
How is this different from an SEO or AI-visibility content agency?
Content agencies write pages and hope crawlers pick them up. We build the interfaces agents actually use: structured product feeds, MCP endpoints, and Claude Skills. Content is the garnish. The feed is the meal.
What exactly does the £10k onboarding cover?
Two months of work: an AI visibility audit across ChatGPT, Claude, Perplexity and Gemini, a structured ACP/UCP product feed, a Claude Skill and MCP endpoint, Agentic Storefront and platform toggles configured to our spec, on-site schema fixes, and a verified before and after report on recommendation share and accuracy.
What does the £1k a month retainer cover?
A monthly monitoring report re-running your buying queries, feed and Skill updates as your catalogue changes, remediation when an AI surface gets your brand wrong, and protocol watch as the standards shift.
What is Stripe's Agentic Commerce Suite and do we need it?
ACS is Stripe's suite for connecting AI-driven discovery to a brand's live checkout and payments. It is how JD Sports went live in January. Most mid-market brands do not need it on day one. The intelligence layer we build is the prerequisite whenever you do, and we prepare the integration path as part of the build, so adopting ACS becomes a question of timing rather than a second project.
Do we need to be on Shopify?
No. Shopify makes activation fastest because Agentic Storefronts are on by default, but the feed, Skill and MCP endpoint work with any stack that can expose your catalogue.
How do you measure success?
Two numbers: recommendation share, how often you appear when AI answers buying queries in your category, and accuracy, whether what it says about your products is true. Both are measured before we start and verified after we ship.
Find out what AI says about you.
Tell us about your brand and we'll run your category through ChatGPT, Claude, Perplexity and Gemini. A free audit across all four, and you keep the results either way.