DocBento: Comprehensive Self-Hosted AI Document Management System

Overview

DocBento is a commercial, self-hosted document management application built with Python and React. It specializes in organizing documents using Optical Character Recognition (OCR), hybrid search, and an AI assistant that provides answers with precise page citations. The system is designed for teams requiring secure, local document processing with advanced AI capabilities.

Benefits

DocBento offers several key advantages for organizations handling sensitive documents. First, it supports four distinct OCR engines to handle various document types, ensuring high accuracy across different inputs. These include PaddleOCR PP-OCRv6, Tesseract 5, Docling for layout analysis, and Vision AI for handwriting and stamps. The system provides a side-by-side view of the original scan and the extracted text, allowing users to verify the accuracy of the extraction.

The optional AI assistant functions as an agent that searches, opens, and reads documents before generating answers. Every statement in an answer is linked to the exact page and document version that supports it. This allows users to verify claims by opening the cited page. The assistant can summarize documents, compare files, extract data, check obligations and deadlines, perform compliance checks, proofread, analyze numbers with charts, draft content, and translate.

DocBento combines full-text keyword search with semantic vector search across multiple languages. The search results include snippets and page numbers. Crucially, the system enforces access controls at the search level; restricted folders are hidden from search results and are not accessible to the AI assistant, even if they appear in the sidebar.

The system tracks expiry dates and sends reminders. Users can create templates with placeholders that can be filled manually or automatically by AI. The AI can also automatically fill in document summaries, types, tags, expiry dates, languages, reference numbers, custom key fields, and entities.

Use Cases

DocBento is ideal for enterprise environments where data privacy and security are paramount. As a self-hosted solution, data remains on the user's infrastructure. When AI features are enabled, users can choose any OpenAI-compatible provider, including local models, ensuring data does not leave the premises if desired.

Teams can use DocBento to manage large volumes of scanned documents, contracts, and reports. The system is particularly useful for organizations that need to verify the accuracy of AI-generated insights. By providing page citations, users can cross-reference information and ensure the AI is not hallucinating facts.

The application supports role-based access control with four built-in roles, custom roles, and groups. This makes it suitable for organizations with complex permission structures. The audit log with CSV export capabilities and processing queue monitor help administrators track scheduled jobs and maintain compliance.

Pricing

Pricing details are not available in the provided information. DocBento is a commercial product, and users should contact the vendor for specific pricing plans.

Vibes

Public reception and reviews are not available in the provided information. However, the system includes a self-resetting demo mode and sample data for evaluation. Users are encouraged to test the system with difficult inputs, such as table-heavy scans and handwritten documents, to assess its capabilities.

Additional Information

DocBento is built with a robust technology stack including Python 3.12, FastAPI, SQLAlchemy, React 19, TypeScript, and Tailwind CSS. It supports PostgreSQL 17 with pgvector or MariaDB 11.8 with native vectors. The deployment uses Docker Compose with HTTPS overlay.

System requirements include Docker Engine 24+ with Docker Compose v2. The minimum hardware requirements are 2 CPU cores and 4 GB RAM, though 4 cores and 8 GB RAM are recommended. The system requires 10 GB of free disk space. It supports x86-64 or ARM64 Linux, macOS, or Windows with Docker support.

Installation involves unzipping the package, running a setup script to create the environment file, and starting the containers. Users can enable AI features by navigating to Settings and adding an API key for an OpenAI-compatible provider. The package includes complete source code and documentation for easy deployment.