chatbot-build-and-deploylisted

# AI Chatbot Build and Deploy Reference This skill covers the complete arc of building and deploying a production AI chatbot: from defining the purpose and choosing channels, through architecture decisions, data quality requirements, integration patterns, security and compliance, accuracy evaluation, and the five-phase build process used by engineering teams working from discovery to production. The central finding running through all documented deployments: chatbot performance is a function of knowledge quality, organizational clarity, and architecture decisions—not model selection. The AI model is a commodity. The knowledge base is the competitive asset. --- ## Part 1: Before a Line of Code Is Written ### Defining Purpose (Step 1) A chatbot without a defined purpose is a demo. The first decision is what the chatbot is specifically responsible for handling—and what it is explicitly not responsible for. Use cases drive everything else: - **Customer service**: Reduce wait times by X%, deflect Y% of tier-1 support volume - **Lead generation**: Qualify Z leads per month at a lower cost-per-lead than current channels - **Internal automation**: Reduce HR query volume by W% in the first two quarters - **Sales enablement**: Handle 24/7 qualification so leads contacted within 5 minutes are 9x more likely to convert than those contacted after 30 minutes (documented finding on lead response timing) - **Employee onboarding**: Route new hires to accurate answers without burdening