Required stack
필요 기술
ShopifyJSONAPI IntegrationNLPLangChainAI Chatbot DevelopmentDialogflow CXRasaGPTShopify APIFacebook Messenger APIWeb Chat Widget DevelopmentNLU Model TrainingData Annotation
Project brief
프로젝트 내용
I run a sporting-goods e-commerce store and I want an AI chatbot that can act like a knowledgeable sales associate focused on recommending the right outdoor-activity equipment to each visitor. The core task is product recommendation, not general customer support or order processing, so every design decision should optimise for guiding shoppers toward the best-suited tents, backpacks, climbing gear, camping stoves, bikes, and similar items from our existing catalogue.
How I imagine the experience
• A shopper describes where they’re headed, their skill level, budget range, or brand preferences.
• The bot instantly analyses that input, checks our inventory feed, and responds conversationally with personalised gear suggestions, upsell add-ons, and concise feature comparisons.
• If an item is unavailable it offers close alternatives rather than dead-ending the chat.
Technical expectations
• You’re free to build with Dialogflow CX, Rasa, LangChain + GPT, or a comparable NLP stack—as long as it supports rich context, product-attribute filtering, and easy retraining when we introduce new SKUs.
• JSON or webhook integration must pull real-time data (price, stock, specs, images) from our Shopify store.
• Responses need to feel human: natural language, short paragraphs, emojis sparingly, no robotic repetitions.
Deliverables
1. A fully configured AI chatbot deployed to our website and Facebook Messenger.
2. Training dataset and intent/entity schema covering outdoor equipment queries.
3. Setup documentation plus a quick-edit guide so my staff can tweak copy or add new products without touching code.
4. A brief hand-off session (video or live call) walking me through dashboard controls and performance analytics.
Acceptance criteria
• At least 90 % of test conversations end with one or more valid product links.
• Average response time
How I imagine the experience
• A shopper describes where they’re headed, their skill level, budget range, or brand preferences.
• The bot instantly analyses that input, checks our inventory feed, and responds conversationally with personalised gear suggestions, upsell add-ons, and concise feature comparisons.
• If an item is unavailable it offers close alternatives rather than dead-ending the chat.
Technical expectations
• You’re free to build with Dialogflow CX, Rasa, LangChain + GPT, or a comparable NLP stack—as long as it supports rich context, product-attribute filtering, and easy retraining when we introduce new SKUs.
• JSON or webhook integration must pull real-time data (price, stock, specs, images) from our Shopify store.
• Responses need to feel human: natural language, short paragraphs, emojis sparingly, no robotic repetitions.
Deliverables
1. A fully configured AI chatbot deployed to our website and Facebook Messenger.
2. Training dataset and intent/entity schema covering outdoor equipment queries.
3. Setup documentation plus a quick-edit guide so my staff can tweak copy or add new products without touching code.
4. A brief hand-off session (video or live call) walking me through dashboard controls and performance analytics.
Acceptance criteria
• At least 90 % of test conversations end with one or more valid product links.
• Average response time