A Zero-Backend Case Study

Created with Cursor AI (From Inbox Chaos to AI-Augmented Mentorship)

Executive Summary

This is a mentoring site that treats AI as an operating layer. An API route ingests a resume, generates an LLM assisted debrief, and delivers insights to Gmail before every Calendly session, without a database, admin panel or CRM.

1. The problem

X mentors engineers and aspiring TPMs. Each engagement needs resume synthesis, gap analysis, and a focused live conversation.

Spreadsheets stores data and they do not create insight.

Key question: How do we increase prepared mentor time per hour of live conversation without adding ops headcount?

2. Design

Layer	Decision	Why
UI	Next.js 14, premium brand UI	Fast, credible, Vercel-native
Intelligence	Groq (Llama 3.3 70B)	Low latency debrief at submit
Scheduling	Cal embed	Best, easy
Delivery	Resend - gmail	Mentor's real dashboard
Persistence	None on v1	Ship in days; limit liability

3. Solution: one API, three outcomes

Candidate journey:

Discover → submit resume or book on Calendly → mentor receives debrief and the PDF in minutes.

Technical spine:

Resume PDF → text extraction → Groq debrief → email with attachment

4. The AI layer: judgment productized

The prompt encodes mentor-grade feedback: strengths, gaps, three actions, <400 words—not ATS keyword fluff.

Outcome: conversation-ready prep.

5. Impact framework (illustrative)

Metric	Before	After
Prep per Candidate	15-30 min	10 min(review debrief and skim PDF)
Time-to-insight	Hours	Minutes
Infra complexity	CRM & maintenance	Serverless, deployed on Vercel

It is an AI-augmented professional service, deployable in a week, extensible when volume demands Sheets, admin, or Calendly–resume matching.