Book a call
Max Tymoshyn, founder of Norml Studio
Max TymoshynFounder, Norml Studio

An AI support assistant that answers from your WooCommerce store

We connect an AI assistant to your WooCommerce store, so shoppers get answers about products, stock and your policies from current store data. Customers who confirm their order details can ask where an order is. Questions it shouldn't answer, such as refunds or complaints, go to your team with the conversation attached.

A few clients we’ve worked with

99Legal / Kremenchuker Law GroupAstraZenecaCoxitDiet vs DiseaseFlorida seal — Regional Counsel projectMetroplex HomebuyersNasdaq Private MarketOffice of Criminal Conflict and Civil Regional Counsel, Second Region of Florida — Ita M. NeymotinPolaclubThe Farm Soho

What we do

Norml connects an AI support assistant to an existing WooCommerce store: your catalog and policies as its answer sources, a chat window on the storefront, order lookups for verified customers and a handoff to your staff. If your store already uses a chatbot product, we assess it before building anything new.

What the assistant covers

Before anything is connected, we agree on what the assistant may answer, which order details it may show and which questions always go to a person. The assistant can read your store but cannot change it.

Solution
Integrations
Connected systems
WooCommerce + Claude
Data flow
WooCommerce supplies products, variations, prices, stock and order status through its REST API, and your approved pages supply policies and size charts. Product and page text is indexed once per item and re-indexed when it changes, so an update replaces the old entry instead of adding a copy. When a shopper asks about a product, its price and stock are read live. The question and the matching store data go to Claude, and the answer returns to the chat in the shopper's language. Order fields are read only after the customer is verified. Handoff topics send the transcript to your inbox or help desk. WooCommerce stays the record for products and orders, and a failed refresh or an undelivered handoff is flagged for the person you name.
  • A review of the chatbot you already useIf your store already pays for a chatbot product or licenses chatbot code, we start there. We check what its plan and license let you change, whether it can read WooCommerce data through an API, where it stores conversations and how it passes a chat to staff. You get a written recommendation on whether to keep it.
    01
  • Answers from your catalog and policiesShoppers get answers about products, variations, prices and stock from WooCommerce, and about sizes, shipping and returns from the pages you approve. Hidden, draft and private products are left out of the data the assistant can read.
    02
  • Recommendations and product cardsThe assistant compares products by their WooCommerce attributes and suggests only items that are visible and in stock. A suggestion can appear as a card with the image, price and stock status. Simple products go into the cart from the card; products with sizes or colors open their product page, so the shopper chooses a variation first.
    03
  • Conversations in the shopper's languageThe assistant detects the language of each message and replies in it. For right-to-left languages such as Hebrew, we test the chat window layout, product names that mix Hebrew and English, and sizes and prices inside right-to-left text.
    04
  • Order status for verified customersLogged-in customers can ask about their own orders. A guest gives the order number and the email used at checkout and gets an answer only when both belong to the same order. The assistant reads the order status, the shipping method and, where your shipping plugin stores one, the tracking number. Nothing else from the order is shown.
    05
  • Handoff to your support teamTopics you mark for staff, such as payment problems, damaged parcels or legal questions, are handed over. The shopper is told a person will reply, and your team receives the transcript, the customer's details and any order number in the inbox or help desk it already uses. If your chat tool supports live agents, staff can take over the same conversation.
    06
  • Logs and an off switchEvery conversation is logged with the sources behind each answer. Unanswered questions and handoffs appear in one list for the person who owns support. The assistant can be switched off at any time, and your usual contact route stays on the store.
    07
Josiah BallCo-Founder & CEO, HOMEEC

“Norml Studio's professionalism and responsiveness were impressive.”

Clutch review
Lucas SeyhunCEO, The Farm Soho

“They were proactive in suggesting improvements we hadn't considered.”

Clutch review
Gaius CooperDirector, Streamline Pump Solutions

“They delivered what they said they would, on time.”

Clutch review
Emma EstefaClient Success Manager, Metroplex Homebuyers

“They focused on how to make our daily operations easier, not just on building features for the sake of it.”

Clutch review
Bret MillerManager, Door Knock Pro

“They understood that the app needed to be practical for someone sitting in a car, not just look good on a desktop.”

Clutch review
Saima AhmedFounder, GoPetZen

“Their level of dedication to the success of the project and GoPetZen's business as a whole is unique.”

Clutch review

What we need from you

  • Access to the storeAn administrator invitation to the WordPress site and to a staging copy of the store, or permission to create one. The assistant's WooCommerce API keys are created with read-only access during setup.
    01
  • Your policies and size chartsShipping, returns and exchange policies, size charts and any FAQ you already publish, in each language the assistant should answer in.
    02
  • Questions customers already askRecent support emails or chat transcripts with personal details removed. They show which topics the assistant should answer or hand over, and they become the test set before launch.
    03
  • An AI model account in your nameA Claude API account with billing in your name, or admin access to the chatbot product you already use. Model usage is billed to that account.
    04
  • Where handoffs should goThe inbox, help desk or live chat tool your team answers from, the hours it is staffed, and the person who reviews the assistant's answers.
    05
  • Details of your current chatbot, if you have oneIts plan, license terms and admin access, so we can check what it can do before recommending changes.
    06

How the setup runs

  1. [Review the store and its support]

    We go through the catalog, plugins, support inbox and any chatbot you already use, then sort past customer questions into topics the assistant answers and topics it hands over.

    Step 01
  2. [Agree sources and limits]

    We agree on the pages and data the assistant may use, the order fields it may show, the handoff topics and where handed-off chats go.

    Step 02
  3. [Build on a staging copy]

    The assistant is connected to a staging copy of the store with outgoing store emails switched off and test orders for the lookups, so real customers don't see it or get messages during the build.

    Step 03
  4. [Test the conversations]

    We test products with variations, out-of-stock items, a policy question in each language, order lookups with right and wrong details, a refund request, and instructions hidden in product text that try to widen the assistant's access. You review the results before launch.

    Step 04
  5. [Launch and hand over]

    We switch the assistant on in the live store and check the first conversations with your support lead. Your team gets notes on reviewing answers and updating the sources.

    Step 05

The people who do this work.

Meet the whole team

No account layer between you and them. These are the Norml specialists this work runs through.

Max Tymoshyn, founder of Norml Studio

Let's chat

“I've spent 10 years working across development and marketing. I started as a developer, but design, SEO and advertising were already part of the job. That taught me to treat a website as one system: the business, the content, the design and the code have to work together. At Norml, I lead the architecture behind our work and stay involved wherever strategy, design, development and search meet.”

Max TymoshynFounder & Architect

From the Norml blog.

Questions

Can you start from a chatbot we already pay for?
Yes, if its plan and license allow the changes. When the review shows it can do the job, we extend it and keep your current setup. When it can't, we connect a model directly, and you decide when to cancel the old subscription. If the vendor's terms leave a license question open, we ask the vendor in writing before changing anything.
Which AI model does the assistant use?
Claude by default, through an API account in your name. If your current chatbot product runs on another model, we work with that model instead. No custom model is trained: the assistant reads your store data each time a question arrives.
How does the assistant handle Hebrew and other languages?
A shopper can switch languages mid-conversation and the replies follow. Prices, stock and sizes come from WooCommerce whatever the language. Policy answers use your approved version in that language; where a policy exists in one language only, you choose whether the assistant translates it or passes the question to your team.
What stops someone from looking up another customer's order?
When the email doesn't match the order number, the assistant shows nothing and offers to pass the question to your team. Repeated failed attempts from the same browser or network block further lookups, so starting a new chat doesn't reset the count. Every lookup, successful or not, is recorded in the conversation log.
Can the assistant issue refunds or change orders?
No. Refunds, cancellations, address changes and discounts stay with your staff. The assistant collects the order number and the reason for the request, then hands the conversation over, so your team doesn't have to ask for them again.
Will every answer be correct?
No. The assistant is set up to say when your store data and approved pages don't cover a question, rather than guess. Each wrong or missing answer found in testing or in your team's reviews becomes a correction to its sources.
Where are conversations and customer details stored?
Transcripts are kept where you choose, such as your help desk or the chat tool's account, under the retention settings available there. The AI model receives the conversation, the store data needed to answer it and, for a verified customer, the order status, shipping method and tracking number. The model provider's own data terms apply to what it receives, so we review them and your account's retention options with you before launch. The assistant is instructed not to ask for card numbers or passwords, and the conversation tests check that it doesn't.
Can the assistant collect email sign-ups for Mailchimp?
Yes, as an optional addition. When a shopper ticks a consent box in the chat, their email goes to your Mailchimp audience with a tag showing it came from the assistant. Follow-up email sequences, such as abandoned-cart or win-back emails, are separate work.
What falls outside this service?
Building or redesigning the store, writing your support policies, translating the catalog and phone support are separate work. A ready-made chatbot license isn't part of this service either; the assistant is set up on your store and your accounts.

See more of our ecommerce work

Each case study covers the starting point, what we built and what changed after launch.

See all case studies