AI Support Escalation Website
Budget: $250 – $750 USD
I need a full-stack web solution that lets visitors describe a technical support problem and receive instant help through an AI assistant available in both text and voice. The assistant should rely on a hybrid approach—pre-programmed flows for common fixes and a machine-learning model (Dialogflow, Rasa, or a comparable NLP service) for anything that falls outside the scripted paths.
When the AI cannot close the ticket, the conversation must hand off seamlessly to a live service advisor. For that step I want an in-browser video call (WebRTC, Jitsi, or Twilio are all fine) so the user never has to install anything. The hand-off logic should also trigger a dispatch notice in an internal dashboard where advisors can claim or be auto-assigned to the session.
Key deliverables
• Responsive website with secure login for users and advisors
• AI chat widget supporting microphone input, speech-to-text, text-to-speech, and natural language responses
• Rule engine + ML pipeline for technical troubleshooting, with easy admin controls to edit scripted paths and add training data
• Real-time escalation flow: AI → video chat → advisor dashboard with status tracking
• Database and API layer for conversation history, ticket status, and analytics
• Basic reporting page showing resolution rate, average handle time, and AI vs. human hand-offs
Acceptance will be based on:
• A live demo on your staging server showing text and voice interactions that resolve at least three scripted technical issues and one ML-handled issue
• Automatic launch of a WebRTC video session when the AI types “Escalating you to a service advisor now”
• Advisor dashboard displaying live queue updates without refreshing the page
• Clear setup instructions so my team can deploy on our AWS account
Feel free to suggest frameworks you work fastest with—React, Vue, or Svelte on the front end and Node, Django, or Laravel on the back end all fit the brief as long as the final product meets the flow above.
When the AI cannot close the ticket, the conversation must hand off seamlessly to a live service advisor. For that step I want an in-browser video call (WebRTC, Jitsi, or Twilio are all fine) so the user never has to install anything. The hand-off logic should also trigger a dispatch notice in an internal dashboard where advisors can claim or be auto-assigned to the session.
Key deliverables
• Responsive website with secure login for users and advisors
• AI chat widget supporting microphone input, speech-to-text, text-to-speech, and natural language responses
• Rule engine + ML pipeline for technical troubleshooting, with easy admin controls to edit scripted paths and add training data
• Real-time escalation flow: AI → video chat → advisor dashboard with status tracking
• Database and API layer for conversation history, ticket status, and analytics
• Basic reporting page showing resolution rate, average handle time, and AI vs. human hand-offs
Acceptance will be based on:
• A live demo on your staging server showing text and voice interactions that resolve at least three scripted technical issues and one ML-handled issue
• Automatic launch of a WebRTC video session when the AI types “Escalating you to a service advisor now”
• Advisor dashboard displaying live queue updates without refreshing the page
• Clear setup instructions so my team can deploy on our AWS account
Feel free to suggest frameworks you work fastest with—React, Vue, or Svelte on the front end and Node, Django, or Laravel on the back end all fit the brief as long as the final product meets the flow above.