I want to set up an AI-driven chatbot inside my existing customer-service software that does one thing exceptionally well: detect when a visitor needs a human touch and hand the conversation off to the right live agent in real time. The hand-off must preserve the entire chat history and pass key context (visitor name, issue type, urgency flag) so the agent can jump straight to helping without repetitive questions. My support stack already handles live chat; what’s missing is the intelligent front line. Whether you prefer Dialogflow, Microsoft Bot Framework, Rasa, or another engine, I’m fine—as long as the final solution sits natively inside the help-desk platform and doesn’t break our current SLA metrics. Deliverables I need from you: • A fully deployed chatbot integrated with the customer-service software’s API/web widget • Accurate intent and entity model that recognises transfer cues (e.g. “need a person”, “speak to agent”, frustration sentiment) • Seamless routing logic that opens a ticket or live chat session with the correct team queue • Brief hand-over guide so my agents understand the workflow and I can adjust intents later I’ll supply API keys, branding assets, and a short list of common queries for training data as soon as we start.