When we talk to Bulgarian online store owners, we hear the same thing over and over: support eats too much of the day. Questions about sizes, stock, delivery times, return procedures. Most of them are identical — and they need identical answers.
The obvious next question is: can something answer these questions automatically? Not with pre-set options and buttons, but with a normal conversation in Bulgarian?
The answer is yes — and today you can try it directly.
What we built
An AI support agent for Bulgarian online stores. The agent is trained on the specific store's catalog, orders and policies. The customer writes their question in Bulgarian — the agent responds in real time, 24/7.
Examples from real conversations with the demo (all data is fictional for the demonstration):
- Customer: "Do you have this dress in size S and in black?" → The agent checks the catalog and answers concretely about availability.
- Customer: "When will I receive order #1042?" → The agent gives the current status and expected delivery date.
- Customer: "I want to return a product — what's the procedure?" → The agent explains the steps according to the store's policy.
of customer inquiries are resolved autonomously by an AI support agent without human intervention — Intercom Fin 2026, Salesforce Agentforce data.
That means for 100 inquiries per month, only 16–30 need your attention. The rest — the agent handles them.
Why a demo, not just a description
Describing an AI agent is cheap. Any company can write "our agent resolves 80% of inquiries." We'd rather show you what it looks like in practice.
The demo on our site is live — it's not a script, it's not a pre-recorded video. You write whatever you want in Bulgarian and get a real answer from the AI model. The agent is trained on the data of StyleBG — a fictional Bulgarian online clothing store with a real catalog of products, orders and policies.
If the agent tries to step outside its role or gets a question outside the store's scope, you'll see how it politely declines and points to the appropriate action. That's an important characteristic — the agent should not "hallucinate" answers when it doesn't know.
How it applies to a real store
The demo uses a fictional store with fictional data. For a real store, the process looks like this:
- We connect the agent to your Shopify catalog — products, stock, prices update automatically.
- We train it with your policies — return conditions, delivery times, answers to frequently asked questions.
- We integrate it into your site — a chat widget, or directly in Messenger/Viber if you prefer.
- We maintain and update it — it's not a one-off installation, it's an ongoing service. If your policy changes, we update the agent.
The entire process from discovery call to a live agent — 2–3 weeks.
Who it's for
The agent delivers maximum value for stores with:
- 30+ customer inquiries per month
- Repeating questions (sizes, deliveries, returns)
- A founder or small team that answers messages personally
- Shopify or a similar platform (for automatic catalog connection)
If you run the store yourself and spend more than 2 hours a week answering repeating questions — that's a real problem with a real solution.