Multi-Platform Customer Support Chatbot
Budget: £250 – £750 GBP
I need an AI-driven chatbot that can resolve the bulk of our customer support questions without handing everything over to live agents. The assistant must function seamlessly on three fronts: embedded on our website, inside our iOS / Android mobile apps, and through Facebook Messenger. Whichever back-end stack you prefer is fine so long as the final experience feels consistent across those channels and can be centrally maintained.
The bot should understand natural language, recognise intent, pull answers from our existing FAQ or knowledge base, keep conversational context, and gracefully escalate to a human when confidence is low. An easy-to-use dashboard to edit responses, review analytics, and train new intents is essential because our content team will handle ongoing updates.
Deliverables
• End-to-end chatbot implementation deployed on website, mobile apps, and Facebook Messenger
• Admin interface for content updates, training, and usage analytics
• Source code, API keys, and setup documentation that let us host or migrate the solution independently
• Short hand-over session (video or live) showing installation, retraining workflow, and fallback escalation path
Acceptance criteria
The bot answers at least 80 % of our current FAQ test set correctly, hands off the rest to a human agent queue without losing conversation history, and maintains parity in behaviour across all three platforms.
The bot should understand natural language, recognise intent, pull answers from our existing FAQ or knowledge base, keep conversational context, and gracefully escalate to a human when confidence is low. An easy-to-use dashboard to edit responses, review analytics, and train new intents is essential because our content team will handle ongoing updates.
Deliverables
• End-to-end chatbot implementation deployed on website, mobile apps, and Facebook Messenger
• Admin interface for content updates, training, and usage analytics
• Source code, API keys, and setup documentation that let us host or migrate the solution independently
• Short hand-over session (video or live) showing installation, retraining workflow, and fallback escalation path
Acceptance criteria
The bot answers at least 80 % of our current FAQ test set correctly, hands off the rest to a human agent queue without losing conversation history, and maintains parity in behaviour across all three platforms.