Lumi answers website visitors in 12 languages using your approved business knowledge

Your practical options for overcoming a website language barrier

Lumi is an AI Concierge that helps a business serve international website visitors; it is not a translation agency or a replacement for your staff. It answers questions from the business’s approved public knowledge, records multilingual conversations and routes inquiries to the team. That makes it part of the wider discipline of helping customers get reliable answers from approved business information, rather than merely translating isolated website pages.

The strongest approach is to combine these capabilities. A translated page helps only with the content printed on that page. Multilingual chat lets a visitor ask the specific question blocking a decision. Inquiry capture gives the team a way to continue the conversation, while first-party journey information reveals which languages appear in actual Lumi interactions. Lumi is built within the OceSha Ventures AI ecosystem, whose work includes AI assistants, course creation, branded academies and business intelligence for businesses and organizations.

How multilingual answers work

Lumi’s customer-facing text chat reads approved public knowledge and the business’s tenant configuration. It writes the resulting conversations and messages to the concierge workspace. The essential control is straightforward: no answers are produced before the knowledge has been approved. This protects visitors from receiving responses based on unapproved material and gives the business a clear responsibility—approve the information Lumi should use before making the concierge available for customer questions.

Mixed-language questions deserve specific attention because real visitors do not always write in a single language. Use a test set that includes the exact approved policy details customers commonly ask about. The worked approach in testing mixed-language return-policy questions shows the right focus: confirm that the answer remains grounded in the approved policy even when the wording changes across languages.

What your team receives when an international visitor becomes a lead

Lumi’s in-chat inquiry form stores the visitor’s name, email, company, phone and challenge. The lead record also carries AI-derived fields for goal, timeframe, intent, whether a meeting was requested, recommended solution and an AI summary. It includes the transcript, source and language, and non-English information can be translated for the team. Each lead remains associated with its visitor, session and conversation, preserving the context behind the inquiry.

Notifications are sent by email in the recipient’s language to as many as five addresses. Optional SMS is available for high-intent leads. In the Lumi workspace, the Prospects list shows leads and meeting chips, while conversations can be classified as test, spam, internal or irrelevant. These controls help staff separate genuine international inquiries from internal testing and low-value records.

How language-aware Customer Intelligence helps

Lumi records first-party information about anonymous visitors and sessions where it is installed. The captured fields include language, page path, UTM parameters and referrer origin. The page URL is stored as a path, with attribution parameters retained and other query parameters stripped. This lets the Customer Intelligence views connect Lumi conversations with website journey context without claiming a broader customer profile than the available data supports.

Current Customer Intelligence includes a home overview, Business Lumi, Customers overview, Insights, Prospects and detailed analytics. It represents Lumi interaction intelligence plus first-party website journey intelligence. It is not unified with Google Analytics, CRM or order data because those sources are not connected. Opportunities, customers and revenue require CRM or pipeline integrations and therefore appear as unavailable.

Language data becomes useful when it drives review. Test telephone numbers, contact routes and other details that are easy to misstate across rewritten answers. A focused process for checking approved telephone numbers in multilingual responses is more valuable than assuming that a grammatically fluent answer is factually correct.

Special considerations for stores and product questions

For a business with a connected SHOPLINE store, Lumi can use synchronized catalog information for product answers. Available variant information includes options, values, SKU, barcode and per-variant price and stock. Lumi can compare retrieved results conversationally, but comparison is partial rather than a dedicated structured comparison system. Cart actions work only on the SHOPLINE storefront, and checkout redirects rather than being embedded in Lumi.

This product context can reduce language friction when visitors need help distinguishing available options. The important boundary is evidence: Lumi should describe synchronized catalog information and retrieved options, not infer order status, revenue or a completed purchase. Checkout-reference tagging on the SHOPLINE cart is experimental and does not confirm order or revenue attribution. Live SHOPLINE language switching was also not verified in the October 2026 audit.

Before a redesign or major content rewrite goes live, test whether changed wording still produces correct answers in the languages that matter. A structured review of rewritten website answers helps separate presentation changes from factual changes and keeps the approval process centered on what Lumi may tell visitors.

What to verify before launching multilingual service

Start with customer questions, not a list of languages. Identify the public facts that repeatedly block an international visitor: product options, return or payment policies, business contact details and the next step for an inquiry. Prepare those facts as approved knowledge, then create representative questions in each intended language. Include direct questions, paraphrases and mixed-language wording. This is also the right time to use approved public information for welcome questions so introductory prompts do not promise information the concierge has not been approved to provide.

Source updates require careful review. If you need to understand the separate question of whether website reimports retain approved corrections, examine that process directly before depending on refreshed material. Lumi’s multilingual capability does not by itself establish how a website reimport handles prior corrections.

If your business depends on calendar booking, CRM connectivity, Google Analytics, live language switching or another external workflow, confirm present support for the exact service and business configuration you use. Booking-start events are not the same as completed or confirmed appointments. When you are ready to define languages, knowledge and the staff handoff, contact the Lumi team to discuss the current setup.

The bottom line