We're seeking an experienced AI consultant to help design and implement an AI-assisted customer service email drafting system for our Google Workspace (Gmail) inbox using training data from all previous customer service replies.
• *Objective**
Build a system that can analyze incoming customer service emails and generate high-quality draft replies (using historic email data) that are reviewed and approved by a human before sending.
• *Current Setup**
• Google Workspace (Gmail) customer support inbox
• No current helpdesk system (open to recommendations if needed)
• Moderate ticket volume (details can be shared during discussion)
• *What We Need**
1. Architecture design (secure + GDPR-compliant)
2. Method for leveraging historical email threads and policy documents
• Likely RAG-based system (open to recommendations)
3. Draft generation inside Gmail or via lightweight interface
4. Human approval workflow (AI cannot send emails directly)
5. Guardrails for tone, policy compliance, and hallucination prevention
6. Clear documentation of the system
• *Nice to Have**
• Experience integrating with Google Workspace APIs
• Experience building AI systems for support teams
• Knowledge of vector databases / retrieval pipelines
• Understanding of GDPR / EU data considerations
• *Deliverables**
• Proposed architecture
• Implementation plan
• Working prototype
• Documentation
• Optional: ongoing optimisation support
Please include:
• Relevant past projects
• Your proposed technical approach
• Estimated timeline
• Rough budget range
Looking for someone pragmatic who can help us implement this properly and securely, not just experiment.