Project Build Log

Customer Support Copilot

An automated email responder using RAG to fetch CRM context before drafting replies. Designed to reduce manual ticket triaging time by 60%.

Status In Development (Step 3/5)

Tech Stack Python, Flask, OpenAI API

Estimated Impact 15 hrs/week saved

Phase 1: Foundation Oct 12

Configured virtual environment, installed dependencies (`flask`, `openai`, `requests`), and successfully authenticated with the mock CRM API.

$ python3 -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt

Phase 2: Webhook parsing Current

Currently pairing with the AI Tutor to debug a TypeError in the Flask route. The nested dictionary isn't unwrapping as expected.

Tutor Context

TypeError: string indices must be integers

Working on adding robust `json.loads()` validation to handle malformed stringified JSON within the payload.

Phase 3: RAG Integration Pending

Next step: pass the cleanly parsed payload along with fetched CRM context into the OpenAI API to generate the draft reply.