AI Customer Support App
Budget: ₹100 – ₹400 INR
I need an AI-driven customer-support assistant that lives inside its own app and handles routine queries for our users 24/7. The goal is simple: reduce the load on our human agents while giving customers fast, friendly answers about orders, accounts, and general product information.
Here’s the scope as I see it:
• Conversation engine – build or integrate a large-language-model–based backend that can understand intent, pull answers from an FAQ knowledge base, and escalate to a live agent when confidence is low.
• Front-end chat interface – a clean, responsive UI for both mobile and desktop access. Push-notification support for ongoing threads is essential.
• Admin dashboard – view transcripts, tag new intents, retrain the model, and monitor key metrics such as average response time and containment rate.
• Integrations – secure hooks into our existing CRM and ticketing system (REST/GraphQL), plus authentication with our user database.
• Documentation & hand-off – clear setup scripts, env-variable map, and a brief video walkthrough so my in-house team can maintain the codebase.
Deliverable passes when the assistant can correctly answer at least 80 % of the supplied test set and fall back to a human hand-off within three seconds in ambiguous cases.
Preferred stack is Python (FastAPI) or Node.js for the backend and React Native or Flutter for the client, but I’m open if you can justify another toolkit. Experience with OpenAI, Rasa, or similar NLP frameworks will be highly valued.
If this sounds like your wheelhouse, let’s get started.
Here’s the scope as I see it:
• Conversation engine – build or integrate a large-language-model–based backend that can understand intent, pull answers from an FAQ knowledge base, and escalate to a live agent when confidence is low.
• Front-end chat interface – a clean, responsive UI for both mobile and desktop access. Push-notification support for ongoing threads is essential.
• Admin dashboard – view transcripts, tag new intents, retrain the model, and monitor key metrics such as average response time and containment rate.
• Integrations – secure hooks into our existing CRM and ticketing system (REST/GraphQL), plus authentication with our user database.
• Documentation & hand-off – clear setup scripts, env-variable map, and a brief video walkthrough so my in-house team can maintain the codebase.
Deliverable passes when the assistant can correctly answer at least 80 % of the supplied test set and fall back to a human hand-off within three seconds in ambiguous cases.
Preferred stack is Python (FastAPI) or Node.js for the backend and React Native or Flutter for the client, but I’m open if you can justify another toolkit. Experience with OpenAI, Rasa, or similar NLP frameworks will be highly valued.
If this sounds like your wheelhouse, let’s get started.