Can I vibecode Fireflies.ai?

text
Build a searchable call transcript archive and meeting intelligence hub to replace Fireflies.ai.
Requirements and stack: A Node.js pipeline plus a local web app using Express, better-sqlite3, and React (running on localhost).
Ingest engine: Watch a local `~/Recordings/` folder using chokidar, or provide a CLI `calls add <file>`. Transcribe audio locally with whisper.cpp, or use Deepgram/OpenAI Whisper if an API key is provided in .env.
AI Summarization: Summarize each call using an LLM (key in .env). Generate a comprehensive overview, bulleted action items with owners, questions asked, and next steps. Store the summary and full transcript in SQLite.
AskFred Clone (AI Chatbot): Add a conversational interface to the app where users can chat with their transcripts. Allow queries like "What was the decision on the marketing budget?" and have the LLM synthesize an answer citing specific meetings.
Global Search capability: Implement an FTS5 keyword search across every transcript ever recorded. Search results must show matching lines with surrounding context, timestamps, and deep links into the full transcript view.
Topic Tracker & Conversation Intelligence: Create a `topics.json` configuration file holding keywords (e.g., competitor names, pricing, churn). Build an analytics dashboard page that lists every call where each topic surfaced, displaying trends and newest mentions first.
Data portability: Provide a one-click export of any call to Markdown format in `~/Calls/notes/YYYY-MM-DD-<name>.md`.
Privacy first: No accounts, no telemetry, no tracking. Audio and transcripts stay completely local unless the user explicitly opts into a hosted API.
Out of scope: Do not build an auto-joining meeting bot for Zoom/Meet/Teams, calendar routing, real CRM sync, or team sharing features. The user will supply recordings themselves.
Documentation: Provide a thorough README explaining dependencies, optional API keys, how to record call audio locally on various OSes, and full setup instructions.