A basic chatbot follows preset paths; an AI assistant uses your business knowledge to handle real visitor requests

The essential difference is flexibility grounded in business knowledge

That difference becomes important after hours, when a visitor cannot simply ask a staff member to interpret an unusual question. A menu-based chatbot works when the available choices already match the visitor’s intent. If the wording, context or follow-up falls outside those paths, the conversation becomes less useful. An AI concierge is designed for the broader task of understanding the request and finding a grounded answer. This is the wider problem addressed by answering website visitors after hours.

Lumi, the AI Concierge for your website, gives each participating business Customer Lumi: a chat and voice assistant embedded on its website. Customer Lumi answers only from that business’s approved public knowledge. This boundary is central to the product. The assistant’s job is not to improvise policies, prices or services; it is to make the business’s actual information easier for visitors to access through conversation.

A modern AI concierge does more than return an answer

Customer Lumi combines conversation with specific next steps. It can capture a lead and hand a visitor a booking link. For a connected SHOPLINE store, it can also use synchronized catalog information to answer product questions, help the shopper find products, choose a variant, display items and manage the store’s cart. Those functions move the interaction beyond a static question-and-answer exchange without pretending that every visitor action represents a completed commercial result.

The distinction is especially visible when the visitor’s words are ambiguous. A request to “cancel,” for example, could refer to different actions in different business contexts. The useful behavior is to clarify the intended meaning rather than force the visitor into the nearest button. The question whether cancel means ending membership or withdrawing from class illustrates why interpreting context is more valuable than matching a single keyword.

The same principle applies to information that is simply absent. If a visitor asks for a service the business has not documented, conversational fluency does not make the service real. A grounded assistant should stay within the available information and support an appropriate handoff. See the practical distinction in handling undocumented furniture assembly requests after hours.

The merchant workspace makes the assistant operational

The visitor-facing conversation is only one half of Lumi. Merchants work inside a private workspace containing knowledge management, conversations, leads, first-party analytics called Customer Intelligence, a private AI assistant called Business Lumi, team management and billing. This gives the business an operating environment for the assistant rather than leaving it as an isolated chat box on the website.

Customer Lumi and Business Lumi serve different audiences. Customer Lumi appears on the website and works from public business information. Business Lumi is private to the merchant workspace, which can also hold private and internal business knowledge. Keeping those contexts separate is important: website visitors need accurate public answers, while the merchant needs private tools for operating and understanding the service.

Customer Intelligence provides first-party analytics within that workspace. Conversations and captured leads give the merchant material to review without turning preliminary activity into a result it has not yet produced. A booking-start event is not the same as a completed or confirmed appointment. Likewise, experimental checkout reference tagging on a SHOPLINE cart is limited and is not confirmed order or revenue attribution.

Connected tools determine what the assistant can actually do

An AI assistant’s actions are only as real as its current connections. Lumi’s commerce capabilities require the business to connect a SHOPLINE store. With that connection in place, Lumi reads synchronized catalog data and uses the search_products, choose_variant and show_items tools in customer conversations. Product search was verified with a development store, but there is no dedicated product-comparison tool.

This is also why apparently simple questions can require more than a scripted response. A visitor comparing an adviser’s published scope and eligibility needs the assistant to distinguish what the business has stated from what the visitor assumes. Comparing advisers’ published service scope and eligibility is a useful example of a knowledge-led request where precision matters more than a broad, generic answer.

Product and service criteria require the same discipline. An assistant should explain published substitution criteria when the business has supplied them, while avoiding invented eligibility rules. That pattern is explored in comparing published window substitution criteria before contacting installers. The goal is not to answer every imaginable question at any cost; it is to answer supported questions clearly and identify where human confirmation is still needed.

Natural conversation matters most when visitors do not follow a script

Real website conversations rarely arrive as neatly categorized support tickets. Visitors use shorthand, change direction and ask follow-up questions that depend on the preceding exchange. A basic chatbot can serve a narrow, predictable flow well, but each additional branch must be anticipated. A conversational assistant is better suited to interpreting those follow-ups while staying anchored to the business information it has been given.

Mixed-language follow-ups are a good test of whether a system is following the conversation or merely matching isolated phrases. The scenario in testing mixed-language follow-ups about garage service options highlights the practical question: can the assistant maintain the visitor’s context without turning an unclear request into a service claim? Language flexibility is useful only when the underlying answer remains grounded.

Sensitive choices also demand exact wording. If a visitor asks whether a meeting-recording practice can be declined, an assistant must reflect the adviser’s published information instead of giving a generic answer that sounds authoritative. Clarifying whether adviser meeting recording can be declined demonstrates why a business-specific knowledge source is a defining feature, not an optional refinement.

The same rule applies to ownership or transfer requests. A visitor asking who can request a voucher transfer needs the business’s actual policy, not a plausible-sounding industry norm. Who can request a beauty gift voucher transfer provides another example of the type of after-hours question that exposes the limits of a fixed FAQ tree.

For businesses in Orange County, Los Angeles and San Diego, CA

In these busy Southern California markets, service providers, retailers and other local businesses often serve visitors who compare several options online and ask specific questions before making contact. A basic chatbot can handle predictable prompts, but an AI assistant is better suited to natural follow-ups about availability, services, products or next steps. The first practical step is to review the questions your business receives most often, then organize accurate public information about services, policies, hours and contact options. That knowledge gives the assistant a reliable foundation for answering visitors and guiding them toward an appropriate action.

Choose the tool by the job it must perform

Lumi supports subscriptions and trials, with monthly AI credits and optional credit top-ups. A tested optional top-up provides 100 credits for $12. Proposed plan structures, proposed usage accounting and proposed pricing are still planned rather than final, so they are not a sound basis for choosing the product today. Evaluate Lumi on its current conversational, knowledge, lead, workspace and connected SHOPLINE capabilities.

Lumi is a product in the OceSha ecosystem, not a separate legal entity or seller. Billing is handled by OceSha Inc., and the product carries the “Powered by OceSha” relationship. OceSha Ventures builds and operates AI-first solutions including course creation, branded academies, AI assistants such as Lumi and business intelligence for businesses and organizations.

The bottom line