Chat with PDF

The Chat with PDF demo application enables intelligent interaction with document content through a conversational AI interface. Upload PDF files and instantly engage in natural dialogue about their contents, asking questions, requesting summaries, and extracting key information with remarkable accuracy.

Features

Transform how you interact with document content through these powerful capabilities:

Setup

To implement the "Chat with PDF" application in your project, configure these essential backend services:

You'll also need to obtain API keys for both the conversational AI models and the embedding models used for text processing.

AI models

This application leverages two complementary AI model types working together:

1. Large Language Models (LLMs): Provide sophisticated natural language understanding to interpret your questions and generate contextually appropriate responses based on document content.
2. Embedding Models: Convert document text segments into numerical vector representations that enable efficient semantic similarity search and Retrieval-Augmented Generation (RAG).

Configure the providers for the models you wish to use:

For comprehensive configuration details, consult the AI SDK documentation covering provider setup and model selection.

Data persistence

The application stores data related to chats, documents, and embeddings to provide a persistent experience.

Database

Learn more about database services in TurboStarter AI.

Application data is organized within a dedicated PostgreSQL schema named pdf:

• chats: captures essential metadata for each document-specific conversation session.
• messages: stores all user queries and AI responses within conversation threads.
• documents: maintains comprehensive tracking of uploaded PDF files, including filenames and storage locations.
• embeddings: contains text segments extracted from PDFs along with their vector representations (using pgvector's vector data type). To optimize similarity searches critical for RAG processing, the system creates an index (embeddingIndex using HNSW) on the embedding column.

Storage

Learn more about cloud storage services in TurboStarter AI.

The PDF files uploaded by users are securely stored in your configured cloud storage bucket. The path field in the documents table maintains the precise reference to each file's location.

Structure

The "Chat with PDF" feature is architected across the monorepo for optimal organization and code reuse:

Core

The @turbostarter/ai package (packages/ai) contains the essential logic under modules/pdf:

• Comprehensive types, validation schemas, and constants specific to PDF processing
• Advanced document parsing, text segmentation, and embedding generation utilities
• Core API logic for managing conversations, performing RAG-based lookups, and interacting with LLMs
• Database operations for storing and retrieving conversations, documents, and embeddings
• Shared utilities for managing PDF file uploads and downloads

API

The packages/api package defines the backend API endpoints using Hono:

• src/modules/ai/pdf/pdf.router.ts: implements Hono RPC routes for document upload and conversation management, handles input validation, applies middleware (authentication, credit management), and invokes the core functionality from @turbostarter/ai.

Web

The Next.js application (apps/web) delivers an intuitive user interface:

• src/app/[locale]/(apps)/pdf/**: contains the Next.js App Router pages and layouts for the document conversation experience
• src/components/pdf/**: houses reusable React components specific to the PDF interaction UI (document upload, conversation interface, message display)

Mobile

The Expo/React Native application (apps/mobile) provides a native mobile experience:

• src/app/pdf/**: defines the screens for the mobile document conversation interface
• src/components/pdf/**: contains React Native components optimized for mobile document interaction
• API integration: utilizes the same Hono RPC client (packages/api) as the web app for consistent backend communication

This architecture ensures that core AI processing and data handling logic is shared across platforms, while enabling optimized UI implementations tailored to each environment.