Call To Action Brief Generator

A small applied AI proof project that turns a sales, client, or discovery call transcript into decisions, risks, owner routing, next steps, CRM-ready notes, and email drafts.

Built for Forward Deployed Engineer, Applied AI Engineer, AI Product Engineer, Solutions Engineer, and customer-facing AI builder roles.

What It Does

• Parses raw call transcripts
• Extracts explicit decisions
• Detects risks, blockers, and technical constraints
• Routes action items by owner lane
• Drafts client follow-up and internal recap emails
• Produces CRM-ready notes
• Exports markdown, JSON, and CSV
• Runs locally with no paid APIs and no dependencies

Quick Start

npm run dev

Open:

http://localhost:4174

Generate a markdown brief from the sample transcript:

npm run brief

Generate JSON or CSV:

node scripts/generate-brief.mjs data/sample-transcript.txt --format=json
node scripts/generate-brief.mjs data/sample-transcript.txt --format=csv

Run the smoke check:

npm run check

Demo Flow

1. Click Load sample.
2. Click Generate brief.
3. Review decisions and risks.
4. Scan the owner routing board.
5. Copy CRM notes or export the markdown packet.
6. Try another transcript by pasting it into the input panel.

Project Shape

index.html                    Browser app
src/app.js                    UI state and rendering
src/parser.js                 Deterministic transcript parser
src/styles.css                Product UI
data/sample-transcript.txt    CLI demo transcript
data/sample-transcripts.js    Browser demo transcripts
scripts/generate-brief.mjs    CLI brief generator
scripts/serve.mjs             Zero-dependency local server
scripts/check.mjs             Smoke check
docs/product-brief.md         Product and role-positioning brief
docs/app-screenshot.png       Browser screenshot

Why Deterministic First

The point of this proof is the workflow shape, not a hidden model call.

The parser is intentionally inspectable: cue maps, routing rules, due-date detection, CRM formatting, and email drafting are all in src/parser.js. A paid LLM extraction layer could be added later, but the app is useful and demoable without one.

Hiring Signal

This project shows:

• customer discovery translation
• applied AI product judgment
• practical workflow automation
• owner/action routing
• CRM and post-call execution awareness
• clean browser app implementation
• ability to ship a polished proof without external services

Product Brief

See docs/product-brief.md.

Positioning Line

I build the bridge from messy customer calls to working AI and software systems.