AI consultant for an online store: how it sells, broken down with the CLIMAT example
How an AI consultant works with a catalog of thousands of products, matches equipment to the buyer's task, and passes ready-made requests to a manager, using CLIMAT as the example.
A shopper visiting an online climate equipment store rarely knows the model name. They know the task: “keep the bedroom cool and quiet in summer”, “heat a country house without gas”, “a water heater for a family of four”. Standard filters don’t handle that. This is where an AI consultant delivers the most value: it understands the task in plain language, selects products from the catalog and explains why it recommends them. Let’s look at how it works, using our project as an example: the climate hypermarket CLIMAT.
- The shopper describes the task, not the model. The consultant selects products from the catalog and explains the choice.
- It recommends only what is in the catalog; prices and availability come from the feed.
- The owner sees all conversations, leads and costs in the dashboard.
- For a store with a ready product feed, launch takes a few days.
Why a large catalog needs a consultant
CLIMAT has over 3,500 products: air conditioners, heating, ventilation, water heaters, air purifiers and dehumidifiers. For shoppers, that is both an advantage and a problem. The choice is huge, but to make a decision you need to understand capacity, sizing, energy efficiency classes and noise levels. Most people don’t want to study this. They want someone to advise them.
In a regular store, a manager fills this role. But a manager isn’t available at 11:00 p.m., can’t keep up with a dozen chats at once and can’t remember the specs of thousands of models. A shopper who gets no reply within a few minutes goes to whoever answers faster.
What the AI consultant does in the store
The AI consultant is not a script with prepared buttons. It is an assistant that works with the real catalog and the store’s knowledge base. In CLIMAT it:
- selects equipment for the task: room area, budget, noise requirements or winter heating needs;
- shows specific models in stock with prices and product cards, not generic advice;
- explains the choice: why this capacity, why an inverter suits an apartment, which model to pick for a child’s room;
- answers questions about delivery, installation and warranty according to the store’s policies;
- passes a ready request to a manager when the shopper is ready to buy: with name, phone number and a summary of the conversation.
Importantly, the assistant does not invent products or prices. It recommends only what is in the catalog, and the catalog updates automatically along with prices and availability.
How it works under the hood
Each shopper question goes through several steps. First, the system filters out bots and spam so they never reach the AI and cost nothing. Next, the platform core searches the catalog for matching candidates: by category, area, budget and specifications. Only after that does the AI form its answer, and it works strictly with these candidates rather than with its own guesswork.
The assistant’s knowledge consists of three parts:
- Catalog: products, prices and availability from the store’s feed.
- Knowledge base — delivery, payment, warranty, installation, and rules for choosing the right capacity.
- Website pages — the assistant rereads the sections on delivery, payment, and services on its own, so it sees changes on the site without manual updates.
What a conversation looks like
A typical dialogue goes something like this. A customer writes: “I need an air conditioner for an 18 m² bedroom that runs quietly.” The assistant asks whether they need heating in winter and suggests two or three inverter models of the right size with a low indoor unit noise level. For each one, it gives a short explanation of why it fits. If the customer asks, “Is there anything cheaper?”, the assistant shows a new selection within the budget. When the person is ready, it asks for a name and phone number, and the manager receives a lead with the full conversation.
For the customer, it feels like talking to an experienced salesperson. For the store, it’s a manager who works around the clock and without a queue.
What else CLIMAT got, besides the consultant
An AI consultant delivers the most value when the store itself is easy to use. That’s why we built a complete set of features into the project:
- automatic product uploads from suppliers — a catalog of thousands of items doesn’t have to be filled in by hand;
- SEO texts and selection tips for every category — so pages get found in search and customers get useful information right away;
- a bilingual website — and the assistant replies in the language the customer writes in;
- an owner’s dashboard showing all conversations, leads, and AI spending.
For more on store development, see our page on turnkey online stores, and for the assistant’s capabilities, see the AI assistant page, where you can chat with a demo version built on the same HVAC equipment catalog.
The dashboard: what the store owner sees
The owner shouldn’t have to guess what the assistant is saying. The dashboard shows:
- all conversations, marked by whether the AI answered or the reply came from the cache of standard answers;
- leads with contact details and the core of the customer’s need, which go straight to the manager in Telegram;
- customer ratings of answers and the option to correct an answer, so the assistant responds correctly next time;
- balance and daily spending, a monthly limit, and alerts when the budget is approaching its cap.
If a conversation gets complicated, a manager can take it over in live chat. If the manager doesn’t reply in time, the system sends a reminder and hands the conversation back to the AI so the customer doesn’t have to wait.
Who this consultant is for
An AI consultant has the biggest impact where products need explaining:
- HVAC and home appliances, tools, building materials;
- stores with large catalogs where choosing without guidance is hard;
- products chosen by parameters: area, power, compatibility, size;
- stores that get many inquiries in the evenings and on weekends.
If you’re still planning your store, start with our article on how to create an online store, and for the difference between a consultant and a regular bot, see our piece AI consultant for a website.
How to launch: four steps
- We define the task — what the assistant should do, which questions customers ask, and where leads should go.
- We connect the data — the catalog feed, the knowledge base, website pages, tone, and rules.
- Test drive — you ask difficult questions, and we refine the answers before launch.
- Launch and ongoing training — the assistant goes live on your site, and then we expand its knowledge based on real conversations.
Frequently asked questions
Won’t the AI make up products and prices?
No. The assistant recommends only products from the store’s catalog, and it takes prices and availability from the feed, which updates automatically. If a suitable product isn’t available, it asks a clarifying question instead of inventing a model.
Will the consultant replace managers?
It takes over routine questions and initial selection. Managers receive warm leads with context and have more time for complex orders.
How does the assistant learn about price and range changes?
The catalog pulls from the feed several times a day, and the assistant rereads website pages on its own. Manual updates are only needed for new rules or promotions that aren’t yet published on the site.
Which languages does it support?
Ukrainian, Russian, and English, in whichever language the customer uses.
How long does launch take?
For a store with a ready feed, it takes a few days, with the longest part being the knowledge base preparation and the test drive.
Want a consultant like this for your store? Tell us about your catalog and goals, and we’ll prepare a demo using your products. Call 097 400 02 22 or message us on Telegram or Viber; all contacts are on our contacts page.
