AI Support Agents

The Problem: As EMR scales, manual support becomes a bottleneck for development. Users need instant answers, but giving everyone access to the full documentation poses a security and clarity risk (e.g., end-users seeing agency-level strategies or partner commissions).

The Solution: Implement a native AI Support system based on RAG (Retrieval-Augmented Generation) with three distinct permission levels. This ensures the right person gets the right information without overwhelming Mannie.

Proposed Structure:

1. Agency Agent (Full Access):

   • Scope: Entire documentation (technical setup, API, agency billing, marketing & commercial strategies).

   • Audience: Agency owners only.
     

2. Sales Partner Agent (The Middle-man):

   • Scope: Partner-specific documentation, sales scripts, commission structures, and lead management tools.

   • Goal: Make partners autonomous in their sales process without exposing sensitive agency-owner data.
     

3. Customer Agent (End-User):

   • Scope: "How-to" guides and end-user features only.

   • Key Feature: Should be "white-labelable" or customizable, allowing us (Agencies) to train it with our own specific data to support our direct clients.

Technical Requirements & Benefits:

• Multimodality: Integration with models like Gemini to allow users to upload screenshots for instant visual troubleshooting.

• Data Privacy: Strict segmentation of sources to prevent information leakage between tiers.

• Efficiency: Drastically reduces "noise" in support tickets, allowing the dev team to focus 100% on the roadmap.
  

Conclusion: By moving from static docs to tiered AI agents, we empower every role in the EMR ecosystem to solve their own problems instantly.