AI engineering · engagement model

AI engineering services for Shopify Plus

We work with £1m+ Shopify Plus operators on fixed-scope, fixed-price AI engineering engagements. 8 to 12 weeks, embedded with your team, shipping production tooling — custom Claude agents, MCP servers, Claude Code skills — that runs in your environment and that you own when we leave. No retainers. No black boxes.Start a conversationFor the broader story of what we do and why, see the AI engineering page.

What an engagement looks like

Weeks 1 to 2 — embed and scope. We get full read access to your Shopify Admin, GA, GSC, and any internal systems relevant to the brief. We sit in your Slack or Teams. By the end of week two there’s a written scope: which workflows are getting automated, what the success criteria are, what we’ll have shipped by the end of week 12.Weeks 3 to 10 — build and ship. Most weeks something goes live. The first system is usually live by week 4 — a working agent or piece of automation doing real work on real data inside your environment. Weekly demos, not monthly status reports.Weeks 11 to 12 — hand off and document. Documentation. Runbooks. Architecture diagrams. Training your team to operate the systems. After week 12, you have production AI systems running in your business that you understand and own.

Three engagement shapes

Most briefs land in one of three shapes. We’ll tell you during scoping which one fits.

Operations agent build

For stores with recurring operational work that humans currently do by hand. Variant-level merchandising decisions, content production for collections and product pages, customer service classification and routing, document and email drafting from templates. We build a custom Claude agent (or several) that handles the recurring work. Typical outcome: work that used to take a person 4 to 8 hours per week now takes them 30 to 60 minutes of review.

Catalogue automation engine

For stores with large or complex catalogues where bulk operations on variants, products, metafields, or content are a recurring need. Cost updates, price rule application, redirect imports, metafield writes, attribute normalisation, product creation from supplier specs. We build a Python-based automation engine against the Shopify Admin GraphQL API that handles batching, rate limits, error retry, and verification. Typical outcome: catalogue operations that previously took multiple days now run in hours.

Custom AI app

For stores that need AI capability woven into a customer-facing or operations-facing application. Custom product configurators with LLM-driven logic, internal dashboards that summarise store state, customer support tooling integrated with order data, finance and remittance automation. We build the application end-to-end — typically Next.js plus Vercel plus Anthropic Claude — designed and shipped to fit the specific use case.

What you get at the end

Production code running in your environment, owned by you
Source code in your GitHub (or wherever you want it)
Documentation covering architecture, prompts, schemas, runbooks
Training for your team — typically 1 to 2 sessions covering operation and basic extension
Metrics showing what shipped, what it replaced, and what it’s now doing

What you bring

System access. Read access to Shopify Admin, GSC, GA, and any internal systems. Technical setup is typically half a day.
A technical counterpart. Someone internal who can answer questions about the business and act as the named decision-maker on scope items.
Willingness to commit. The engagement is a focused 12-week sprint. Stores that try to fit it in around five other priorities tend to get less out of it.

Pricing and timelines

Single agent or focused automation6 to 8 weeks£20,000 to £40,000

Standard 8 to 12 week engagement8 to 12 weeks£30,000 to £60,000

Larger or multi-system build12 to 16 weeks£60,000 to £120,000

Custom AI app (full application build)12 to 16 weeks£40,000 to £100,000

All prices in GBP, excluding VAT. Payment typically split 50/50 — half on engagement start, half on hand-off. Some clients prefer monthly milestones; we can do that.

Is this a fit?

Three honest checks. If yes to all three, the conversation is worth having.

Do you have at least one workflow that, if AI took it over, would meaningfully change your operating model?

Not "would be slightly faster". Meaningfully. Variant-level merchandising across thousands of SKUs, content generation across hundreds of pages, customer support classification at high volume, document creation that fills your team's week. If you can't name one, an engagement is premature.

Are you operating at a scale that justifies the budget?

£30k+ is real money. Stores doing £1m+ in annual revenue with the right kind of operational complexity see compounding return on the work; smaller stores often don't. We'll give you an honest read on the call.

Do you have a technical counterpart on your side?

Doesn't need to be a developer — needs to be someone who knows the business and can make scope calls. Engagements without a clear counterpart struggle.

If yes to all three, start a conversation. If no to one of them, we’ll usually have a useful steer on what to do instead.

Frequently asked questions

How is this different from an "AI agency" retainer?

Most AI agency retainers bolt one or two GPT prompts onto an existing service ("AI-powered SEO", "AI content automation") and bill monthly. The work doesn't compound and there's no software you own at the end. An engagement here is a fixed-scope build that produces real software running in your environment. After the engagement ends, the systems continue running without paying us anything.

Do we own everything you build?

Yes. Code, prompts, schemas, MCP server configurations, documentation. We give you everything in a GitHub repo of your choice. We don't retain rights, we don't paywall extension, we don't hold your systems hostage.

What if our team can't operate the systems after handover?

Part of the engagement is training your team. If after handover you find you need ongoing support, we offer optional ongoing optimisation arrangements — but it's never a lock-in, and most clients don't need it. The systems are designed to run without us.

Can you work with our existing developers?

Often the better arrangement. We work alongside in-house teams regularly — we ship the AI engineering layer, your team handles the rest. We can also work with external dev agencies if that's how your stack is set up.

How do you handle data privacy?

We sign your standard NDA before any system access. We use your Anthropic API account (or set one up for you) so all data stays in your account, not ours. We follow your data handling policies, including any UK GDPR or US state-level privacy requirements. Specifics agreed during discovery.

Can you work with non-Shopify stores?

Most of our work is Shopify Plus, but we've shipped non-Shopify builds — the property finance operations platform is the public example. The engineering pattern works wherever there's structured operational work that's currently done by hand. If you're on WooCommerce, BigCommerce, custom commerce, or a non-commerce business with similar shape, the conversation is still worth having.

How fast can we start?

Typically 4 to 6 weeks from a first conversation to engagement start, depending on scheduling. For urgent work, sometimes faster. If your timeline is tight, mention it on the call.

Start a conversation

30 minutes. No pitch. Tell us what you’re trying to automate or build. We’ll either tell you yes we can help and here’s what an engagement looks like for your shape, or no this isn’t a fit and here’s what we’d suggest instead.Talk to us