I need an AI voice agent that can pick up incoming phone calls and handle customer-support requests focused on general information—opening hours, order status look-ups, service availability, and similar FAQs. Here’s the high-level flow I’m aiming for: when a customer dials our support number, the agent greets them, identifies intent through natural language understanding, fetches the requested details which delivers the answer with natural-sounding text-to-speech, then offers to transfer to a live rep only if necessary. I’m open to Dialogflow CX, Amazon Lex, or an on-prem solution, but whichever stack you choose must integrate smoothly with Twilio voice (SIP trunk is fine) so we can keep our current phone provider. Whisper-quality speech recognition, real-time sentiment detection, and analytics dashboards for call logs will be important for us to iterate. Please ensure fallback phrases and error handling are polished; the caller should never feel “stuck.” Deliverables: • Conversation design documentation with intents, entities, and sample utterances • Working voice bot deployed to our Twilio number in a staging environment, with full source code • Connection scripts that query our API and return JSON responses to the bot • A brief hand-off guide so our internal team can retrain or update intents later Acceptance criteria will be a test suite of twenty common inquiries where the agent answers correctly, plus less than three seconds average response latency. If this pilot performs well, we’ll expand to billing and technical support flows next quarter.