Mistral LLM Website Support Chatbot
Budget: ₹12,500 – ₹37,500 INR
I need an AI chat assistant built around Mistral’s large-language-model and exposed through a clean, well-documented API that drops straight into my website. Its sole purpose is customer support, so every feature should focus on resolving technical questions and everyday “how-do-I?” enquiries without human intervention unless absolutely necessary.
The workflow I have in mind is simple: a visitor opens the chat widget, the bot pulls the relevant context from our knowledge base, delivers an accurate answer, and—if confidence drops below an agreed threshold—hands the session to a live agent while passing the full conversation log. Low latency, clear fallback logic, and rock-solid data privacy are critical.
Please include:
• Production-ready REST or GraphQL endpoints backed by the Mistral LLM
• A lightweight, embeddable JavaScript widget styled to match our site
• Prompt design or fine-tuning routines so the bot understands both technical support and general enquiries
• Conversation logging, basic analytics, and an admin toggle to update FAQs on the fly
• Deployment scripts, environment variables template, and a short hand-over call or video guide
The project is finished when I can paste the widget into my staging site, run a few sample questions, see correct answers returned from the live API, and switch it to production with a single config change.
The workflow I have in mind is simple: a visitor opens the chat widget, the bot pulls the relevant context from our knowledge base, delivers an accurate answer, and—if confidence drops below an agreed threshold—hands the session to a live agent while passing the full conversation log. Low latency, clear fallback logic, and rock-solid data privacy are critical.
Please include:
• Production-ready REST or GraphQL endpoints backed by the Mistral LLM
• A lightweight, embeddable JavaScript widget styled to match our site
• Prompt design or fine-tuning routines so the bot understands both technical support and general enquiries
• Conversation logging, basic analytics, and an admin toggle to update FAQs on the fly
• Deployment scripts, environment variables template, and a short hand-over call or video guide
The project is finished when I can paste the widget into my staging site, run a few sample questions, see correct answers returned from the live API, and switch it to production with a single config change.