Offre de lancement : Accès à vie à 79€ (code MZMDKXNW, 30 premiers clients)·En profiter ➔
← Back to Blog

HOW TO BUILD A PRODUCTION READY RAG AI

Published on 2026-11-09

2 free analyses · No credit card · Results in 30s
🚀

HOW TO BUILD A PRODUCTION READY RAG AI

Source Title :
How To Build A Production Ready RAG AI - My Full AI Coding Workflow
Source URL :
Scan Date :
09/11/2026 at 08:03
Generator :
VidScope Agent (Autonomous Mode)
📄

EXECUTIVE SUMMARY

This master class details the comprehensive development of a production-ready Retrieval Augmented Generation (RAG) system designed to index and search an entire library of blog posts and YouTube videos. The project covers the full development lifecycle, including database schema management with Drizzle ORM, secure authentication, and robust ingestion workflows using automated cron jobs. Throughout the series, the developer focuses on modularizing business logic, implementing semantic search, and refining code for performance and security. The final stages emphasize troubleshooting deployment, fixing vulnerabilities, and establishing reliable CI/CD pipelines to ensure a functional and scalable AI-powered search application.

Tool Name Category Quick Description Official Link
Neon
Sponsor
Database
Serverless PostgreSQL database with branching support used for vector search capabilities.
« We're going to be using Neon for our database of choice. »
Visit
Next.js
Used
Framework
The frontend and backend framework used to build the web application.
« We're also going to be hosting on Verscell since we're writing in Nex.js »
Visit
CodeRabbit
Used
Dev tool
AI-powered tool used for automating code reviews and ensuring code quality within the repository.
« we're going to be using Code Rabbit for everything for our code reviews. »
Visit
GitHub
Used
Dev tool
Platform for version control and hosting project repositories with branch protection rules.
« I actually want to start with a GitHub repository since when I create real world projects, the first thing I do is create a repository for them. »
Visit
Visual Studio Code
Used
IDE
Code editor used for writing, testing, and managing the project files.
« And now we're going to open up just essentially a blank VS Code terminal. »
Visit
Docker
Used
Dev tool
Containerization tool used for sandboxing the AI environment to improve security and ease of use.
« I use Docker sandbox for it. »
Visit
Drizzle ORM
Used
Database
TypeScript ORM library used for managing database schemas and queries.
« I'm going to be using the newest version of Drizzle, the Drizzle 1.0 release. »
Visit
T3 Env
Used
Dev tool
Library used for managing and validating typed environment variables.
« We're going to be using T3 env for that. »
Visit
TypeScript
Used
Language
Used for defining database schemas, relations, and configuration files.
« So I can create a new file called schemas.ts. »
Visit
Drizzle Kit
Used
Dev tool
CLI tool utilized to manage migrations, generate migration files, and push schema changes to the database.
« We need a drizzle config for that. »
Visit
Better-Auth
Used
Authentication
Authentication library used to handle user login via GitHub.
« So, I can say npmi better off. »
Visit
Drizzle
Used
Database
Used as the ORM to manage database schemas, migrations, and schema configuration.
« inside of our Drizzle config, we can actually get rid of all the changes »
Visit
Vercel Workflows
Used
Also mentioned as: Vercel AI SDK
Cloud
Platform feature used to orchestrate ingestion pipelines for blog and video content.
« we're going to be using those Verscell workflows to ingest all of our data. »
Visit
Cheerio
Used
Dev / Web Scraping
Library used for parsing HTML content during the article ingestion process.
« For example, this Cheerio library that it's using, we want to make sure we get imported. »
Visit
PostgreSQL
Used
Database
The primary relational database used for storing application data, including vectors and UUIDs.
« We're going to say UU ID coming from Postgress. »
Visit
RSS Parser
Used
Library
Used to fetch and parse RSS feeds for blog content ingestion.
Visit
Zod
Used
Validation Library
Used to create schemas and validate the structure of RSS feed items to ensure data integrity.
« Make sure we use zod for that. »
Visit
LM Studio
Used
Local LLM Platform
Used to host and run local AI models, specifically for embedding tasks during development.
« So, what I'm actually going to do is inside of LM Studio, which is where I run all my local models, I'm actually going to load a local embedding model. »
Visit
OpenAI
Mentioned
AI Provider
Mentioned as the target provider for production-grade embedding and AI tasks.
« You could go ahead and you could use like OpenAI for the embedding, which is what we will use in production. »
Visit
Git
Used
Version Control
Used for managing code branches and committing changes to the article ingestion workflow.
« So, we can say checkout-b and this is going to be ingest articles. »
Visit
REST Client
Used
Dev tool
VS Code extension used to perform HTTP requests to test ingestion workflows locally.
« This extension is called rest client. »
Visit
npm
Used
Dev tool
Package manager used to run the development environment.
« npm rundev. Just like that. »
Visit
YouTube Data API v3
Used
API
Utilized to fetch channel content, captions, and video details for the RAG ingestion process.
« you'll see YouTube data API v3. »
Visit
Google Cloud Console
Used
Cloud
Platform used to manage API keys, OAuth credentials, and enable services for the YouTube API.
« So, what we need to do is we need to go to the Google Cloud Console API. »
Visit
🔍

UNRESOLVED MENTIONS

Tools or technologies mentioned vaguely or without an exact name in the video:

better off ⏱️ 14:02
A library for authentication mentioned as being set up after Drizzle.
grill me skill from Matt PCO ⏱️ 11:51
An AI command/skill used to prompt the AI to ask questions about the project.
local AI ⏱️ 27:28
The author mentions using a local model for AI-assisted coding instead of a larger frontier model.
my AI ⏱️ 40:10
The speaker uses an integrated AI coding assistant to refactor code and perform automated tasks throughout the workflow.
frontier model ⏱️ 55:58
A reference to high-end, heavy AI models running in the cloud versus a local implementation.
A local embedding model used for development purposes alongside OpenAI.
AI sandbox ⏱️ 01:58:22
An environment used to generate UI code and components via prompts.
better off secret ⏱️ 02:01:49
A secret generation method used for securing cron jobs during local testing.
subtitles npm package ⏱️ 02:19:47
A library mentioned for parsing captions from YouTube videos.
Google API's npm package ⏱️ 02:19:27
A package suggested to the AI for handling YouTube API interactions.
caption parsing library ⏱️ 02:39:15
A library used to download and parse YouTube captions as SRT files.
embedding library ⏱️ 02:40:55
A library invoked to perform embedding on the processed text chunks.
⏱️

KEY MOMENTS

[00:28]
Overview of the production workflow and agent-based coding strategy.
[11:15]
Using an AI sandbox to flesh out architectural decisions and requirements.
[23:06]
Executing database migrations and pushing schema changes to the live environment.
[41:16]
Configuring security environment variables and generating secrets for Better Auth.
[57:46]
Moving chunking logic into its own modular library file for better maintainability.
[01:12:32]
Building a generic batch execution utility to handle asynchronous operations efficiently.
[01:30:03]
Setting up real-world test cases to verify the chunking logic on actual blog content.
[01:47:16]
Configuring 'vercel.json' to set up automated cron jobs for weekly data ingestion.
[01:59:50]
Securing cron job endpoints using a secret header validation pattern.
[02:18:44]
Utilizing Code Rabbit AI to perform automated code reviews on pull requests.
[02:36:12]
Setting up the video ingestion logic to handle YouTube playlist items.
[02:54:40]
Generating database migrations to update the schema for new user search functionality.
⭐

TECHNOLOGIES DISCUSSED

🔹 RAG
🔹 TypeScript
🔹 PostgreSQL
🔹 CI/CD
🔹 Next.js
🔹 RSS
🔹 Vector Database
🔹 AI Agents
🔹 OAuth
🔹 ORM

Ready to save hours every week?

Let VidScope's AI analyze your YouTube videos and extract the best tech tools instantly.

2 free analyses · No credit card · Results in 30s