AI Customer Support Platform
Budget: $30 – $250 USD
I need an AI-driven platform whose sole goal is to handle customer support queries. The system must respond naturally and accurately across two initial channels:
• Website chat
• Email
Core expectations
The assistant should ingest a knowledge base that I will supply (HTML / PDF / CSV) and use it to answer routine questions, route complex issues to a live agent, and keep a conversation transcript for reporting. A simple web dashboard will let me update articles, review conversations, and train new intents without touching code. If you have a preferred stack—Python with FastAPI, Node.js, or a no-code framework—note it; I care more about reliability and model quality than language choice.
Integration & scalability
Please architect with popular LLM APIs or an open-source model you can fine-tune so we avoid vendor lock-in. Web chat should drop in via a lightweight JavaScript widget, while email support can connect through standard IMAP/SMTP or a service like SendGrid. Social media support may be added later, so design with modularity in mind.
What I’d like to see from you
Share past work that proves you have built or integrated conversational AI, chatbots, or ticketing automation before. Links to live demos, GitHub repos, or short video walkthroughs are perfect. A concise outline of how you would tackle data privacy and latency will strengthen your proposal.
Deliverables
1. Deployed MVP (chat widget, email listener, admin dashboard)
2. Source code in a private repo
3. Setup and hand-off documentation
4. One brief training session via video call
Acceptance criteria
• First-response accuracy ≥ 85 % on a test set I provide
• Average reply time < 2 s for chat, < 30 s for email polling
• Zero critical errors during a 7-day pilot run
If the pilot succeeds, ongoing optimisation and potential social media channel expansion will follow on new milestones.
• Website chat
Core expectations
The assistant should ingest a knowledge base that I will supply (HTML / PDF / CSV) and use it to answer routine questions, route complex issues to a live agent, and keep a conversation transcript for reporting. A simple web dashboard will let me update articles, review conversations, and train new intents without touching code. If you have a preferred stack—Python with FastAPI, Node.js, or a no-code framework—note it; I care more about reliability and model quality than language choice.
Integration & scalability
Please architect with popular LLM APIs or an open-source model you can fine-tune so we avoid vendor lock-in. Web chat should drop in via a lightweight JavaScript widget, while email support can connect through standard IMAP/SMTP or a service like SendGrid. Social media support may be added later, so design with modularity in mind.
What I’d like to see from you
Share past work that proves you have built or integrated conversational AI, chatbots, or ticketing automation before. Links to live demos, GitHub repos, or short video walkthroughs are perfect. A concise outline of how you would tackle data privacy and latency will strengthen your proposal.
Deliverables
1. Deployed MVP (chat widget, email listener, admin dashboard)
2. Source code in a private repo
3. Setup and hand-off documentation
4. One brief training session via video call
Acceptance criteria
• First-response accuracy ≥ 85 % on a test set I provide
• Average reply time < 2 s for chat, < 30 s for email polling
• Zero critical errors during a 7-day pilot run
If the pilot succeeds, ongoing optimisation and potential social media channel expansion will follow on new milestones.
Related categories:
JavaScript
Python
Django
Machine Learning (ML)
HTML
API Integration
AI Chatbot Development
AI Development