Creation of a chat-bot

Business requirements for the chatbot

1. Integration and visual representation

1.1 Integration form

• The chatbot must be implemented as an iframe widget, embedded on the OpenCart site

• The widget must load asynchronously, without slowing down the main site

1.2 Visual design

• The color scheme of the widget must automatically adapt to the site's branding (primary and accent colors)

• The chat icon should be noticeable but not aggressive (recommended size: 60×60px)

• Upon first visit or after a period of inactivity, a welcome bubble with a call to action should appear (e.g., “Need help choosing equipment?” or “Need a consultation?”)

• The bubble should automatically hide after 5-7 seconds or upon user interaction

Desktop: the widget is fixed in the bottom right

The chat button should not overlap navigation elements, the cart, and other critical interface elements

• Positioning should take into account the bottom navigation panel (if present)

2. Session management and synchronization

2.1 Session Logic

• The system must retain the context of the dialogue within the user's session (collected parameters: sport type, budget, size, age, etc.)

• The dialogue history must be saved for at least 24 hours

• When the user returns, the bot should greet them considering the previous context (“You were looking at futsal shoes. Shall we continue?”)

2.2 Synchronization between tabs

• If the user opens the site in multiple browser tabs, all messages must synchronize in real-time

• When a message is sent in one tab, it should appear in all open tabs

3. Request routing and escalation to a manager

3.1 Triggers for connecting a manager

• Large orders: check from 5 million sums or quantity of items >15 pcs.

• B2B requests: requests for teamwear, personalization (applying logos/numbers)

• Complex questions: the bot could not provide an answer after 2-3 clarification iterations

• User request: explicit request to contact an operator

Contact details: name, phone

• Essence of the request: brief summary of the dialogue

• Selected products: articles, links, quantity

• Priority: high (large check), medium (teamwear), normal

• Timestamp: date and time of the request

4. Functional requirements (based on the existing technical specification)

4.1 Understanding natural language requests

• Recognition in Russian and Uzbek languages

• Support for voice input in Uzbek language

• Extraction of parameters: sport type, gender/age, budget, season, type of surface

4.2 Smart product selection

• Integration with the OpenCart catalog (categories, attributes, availability, prices)

• Ranking: 3-5 most relevant products considering popularity, margin, availability

• Enrichment of product cards: technologies (MICRO-MESH, BRAMA), reviews, sizes

4.3 Assistance with size and sets

• Size recommendations based on historical data (“this model runs small”)

• Auto-selection of sets: for shoes → socks, shin guards, bag

• Calculation of total amount and savings

4.4 B2B and personalization

• Gathering requirements for teamwear: number of players, colors, logos, budget

• Cart calculation considering the application of numbers/logos

• Automatic escalation of large orders to a manager

4.5 AI services

• Visual selection by photo (Fashn.ai)

• Virtual fitting

• Integration with the Tilmoch AI model for translating responses into Uzbek

• Consider voice responses

5. Analytics and optimization

5.1 Logging

• Reasons for refusals and unfinished sessions

• Frequently asked questions and requests that the bot could not answer