PDF ChatBot powered by Next.js, LangChain, and GPT4

An open-source AI chatbot to chat with multiple PDF files.

Tech Stack · Running Enviroment · Deployment · Run the server · References

An open-source AI chatbot to chat with multiple PDF files.

Tech Stack

• Next.js 13 App Router
• Typescript as a language
• Tailwind CSS as a CSS framework, with light and dark themes
• NextAuth.js for authentication
• Powered by LangChain
• OpenAI gpt-3.5-turbo as a language model (You can change it to gpt-4 if you have access to it)
• Pinecone as a vectorstore
• PostgreSQL and Sequelize as database and ORM for managing users and chat history

Running Enviroment

Node: 18.x

Deployment

1. Clone the repo or download the ZIP

git clone [github https url]

2. Install packages

npm install

3. Set up .env.development

• Copy .example.env.development to .env.development for local dev

• Copy .example.env.production to .env.production for production

• Fill env variables

OPENAI_API_KEY=
PINECONE_API_KEY=
PINECONE_ENVIRONMENT=
NEXTAUTH_SECRET=

• Get an API key on openai dashboard and fill it in OPENAI_API_KEY.
• Create an API key on pinecone dashboard and copy API key and Environment and then fill them in PINECONE_API_KEY and PINECONE_ENVIRONMENT.
• Generate a random 32-digit hexadecimal string and fill it in NEXTAUTH_SECRET.

4. Set up database.

• Create a database in PostgreSQL on your local or remote For local dev, In sequelize/config/config.js, replace variables with your correct values under development. For production, In sequelize/config/config.js, replace variables with your correct values under production.

5. In lib/server/makechain.ts chain change the QA_PROMPT for your own usecase. Change modelName in new OpenAI to gpt-4, if you have access to gpt-4 api.

6. In the Starter(free) plan, you cannot delete saved vectors in an index. In non-starter plan, you should open these lines

• line 71 in lib/server/embeddings
• line 115 in app/api/file/route.ts

Run the server

• Run the app npm run dev to launch the local dev environment

npm run dev

Then this should be running on localhost:3000

• Upload PDF files and then type questions

References

The parts of Chain and Embedding were inspired by GPT-4 & LangChain