Messy PDFs, slide decks, spreadsheets, and web pages become clean, structured text for AI pipelines once Docling Serve has converted them. Open the built-in playground, paste a link or drop a file, and press Convert to see the result beside the options that produced it. A page viewer draws the detected layout over each scanned page, with filters for text, tables, and pictures. Output tabs show Markdown, HTML, and JSON, and one click downloads everything as a zip. The options panel picks OCR engines, table structure models, picture descriptions, code and formula enrichment, and chart extraction, each with its own preset. Copy the exact request as a curl command or a Python snippet and move it straight into production code. A Server tab lists the limits, supported sources and targets, installed library versions, and every model preset the instance offers. Optional API keys, S3 or Google Drive sources, and async task queues cover shared team deployments. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

With 2,600+ GitHub stars since its May 2026 open-source launch, Knowhere solves the last-mile problem of document intelligence for AI systems — transforming complex unstructured PDFs, reports, and multi-page documents into structured JSON chunks that LLMs can consume without hallucination. The platform processes documents through an AI-native parsing pipeline that handles 20+ page documents with deep hierarchies, intricate tables, and multimodal content including images with OCR, achieving 95% precision in information extraction while reducing token costs by 50% compared to raw document ingestion. The knowledge tree architecture maintains historical context across multiple documents, enabling cross-document graph navigation for agentic retrieval that goes beyond simple chunk-based RAG. Built on Python 3.11+ with MinerU as the default PDF parser, the backend API runs alongside async workers that process document ingestion, graph construction, and embedding generation. The self-hosted Docker Compose stack packages the API server, processing workers, and Next.js dashboard for managing API keys, webhooks, and document-processing jobs, backed by PostgreSQL and Redis. Both Python and Node.js SDKs provide programmatic access for integration into existing AI pipelines and agent frameworks. LLM providers include DeepSeek and Alibaba Cloud DashScope with configurable key rotation for rate-limit management. Deploy on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache 2.0 licensed.

Docling Studio shows you exactly what a document parser sees, drawing a colored box around every title, paragraph, table, and picture it extracts from your PDFs. Import a PDF, run an analysis, and the result opens with three linked panels: the document tree on the left, the paged preview in the center, and the selected element on the right with its type, page, position, and text or table. Click any box on the page and the same element lights up in the tree and properties panel as other boxes fade away. The layers bar counts each element type and toggles boxes per type, or hides them all to reveal the bare page. You can analyze several documents at once from the list, watch progress percentages on long files, and queue work three analyses at a time. Ask a question and a local model answers from the document, with a reasoning timeline that links each step back to the passage it read. Finished results download as the original PDF, clean Markdown, or Docling JSON for downstream pipelines. AI agents connect over MCP to read and cite your documents, and an optional connection to your own Docling Serve server handles heavier parsing loads. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Adobe Acrobat and Smallpdf, answered by a self-hosted Java web application: Stirling PDF processes every file with its 60+ tools on your own server and deletes it after the task completes. Nothing is uploaded to a third party, which is the whole point for contracts, invoices, and medical records. The toolbox covers page operations (merge, split at page numbers or scanned dividers, rotate, reorder, crop, extract), conversion in both directions between PDF and Word, Excel, PowerPoint, images, HTML, and Markdown, OCR that turns scans into searchable PDFs via Tesseract/OCRmyPDF (including PDF/A archival conversion), and security tools for passwords, permissions, watermarks, signatures, and true UI-driven text redaction. A built-in viewer handles annotation, drawing, and text or image insertion, and V2 added stateful processing - upload once, chain multiple tools - plus undo/redo history. For automation, nearly every tool has a REST API endpoint, no-code pipelines combine operations into custom logic chains, and watched folders process files automatically. Enterprise deployments get SSO, user management, and audit logging; the interface ships in 40+ languages. With 84K+ GitHub stars it is the most popular PDF tool in self-hosting, replacing $20/month Acrobat subscriptions with flat infrastructure cost.

Send a document, receive a PDF: Gotenberg is a Docker-based, stateless HTTP conversion API trusted in production by thousands of companies and adopted by notable open-source projects. Send files as multipart/form-data, get a PDF back; Chromium, LibreOffice, and fonts are the container's problem, not yours. Headless Chromium handles URLs, HTML templates, and Markdown with pixel-perfect browser rendering: it executes JavaScript, loads web fonts, waits for network idle, a JS expression, or a DOM selector before rendering SPAs, and accepts injected cookies and HTTP headers for authenticated pages. LibreOffice converts 100+ office formats - .docx, .xlsx, .pptx, and legacy formats - with page-range extraction and PDF/A archival conformance. Built-in PDF engines round out the pipeline in every image variant: merge, split, rotate, flatten, encrypt, watermark, stamp, read/write metadata and bookmarks, plus Factur-X/ZUGFeRD e-invoicing and PDF/UA accessibility compliance. Screenshots of URLs and HTML come from the same endpoints. Zero-transfer pipelines stream files directly between S3, MinIO, or GCS presigned URLs and Gotenberg, bypassing your application entirely, and webhooks enable async processing. Statelessness means horizontal scaling is trivial - run as many replicas as your conversion volume demands. Three image variants (full, Chromium-only, LibreOffice-only) trim the footprint to what you actually use.

Receipts, contracts, and letters that arrive as pixels rather than text: Papermerge is a document management system built specifically for scanned documents. Upload a PDF, TIFF, JPEG, or PNG and OCR runs automatically, with a real-time status indicator beside the document title; under the hood it drives OCRmyPDF and Tesseract, supporting 130+ languages and producing a new document version with a selectable, searchable text layer you can download. Full-text search runs across everything, with multiple search-engine backends (Xapian by default). The interface deliberately mimics a modern desktop file browser: dual-panel commander, drag and drop, hierarchical folders, and colored tags on documents and folders. Page management fixes what scanners get wrong - delete blank pages, rotate, reorder, merge, extract, and move strayed pages between documents, powered by PikePDF. Document types act as categories, each with its own custom metadata fields, so invoices carry vendor and amount while contracts carry parties and dates. Versioning preserves every state of a document. Multi-user support includes groups, group ownership, permission management, and document/folder sharing between users and groups, and an OpenAPI-compliant REST API automates ingestion from scanners or scripts. Apache-licensed, ideal for long-term digital archives.

A pile of paper becomes a searchable digital archive under Paperless-ngx, the community-maintained document management system. The pipeline is what makes it work: drop a file into the consumption folder (or let it poll an email inbox), and a worker hashes it for duplicate detection, routes it by type, runs OCRmyPDF with Tesseract (100+ languages) on anything without a text layer, and produces an archivable PDF/A with invisible selectable text embedded - so Ctrl+F works on your scans. Then the smart part: a scikit-learn classifier (TF-IDF plus multi-label prediction) trained on your own tagged documents automatically assigns tags, correspondents, and document types to new arrivals, alongside rule-based exact, keyword, and regex matching for deterministic cases. Full-text search includes relevance ranking, match highlighting, autocomplete, and "more like this" similarity search. Apache Tika integration extends consumption to Word, Excel, PowerPoint, and LibreOffice formats. The Django/Angular app adds custom fields, saved views on a customizable dashboard, bulk editing, workflow automation triggered on upload or matching, shareable public links with expiration, and a robust permissions system with per-document access control. Metadata lives in PostgreSQL with Redis-queued Celery workers processing documents in parallel. Your tax records, invoices, and contracts stay on your server - searchable in seconds, never in someone else's cloud.

RAGFlow has established itself as one of the most widely adopted open-source RAG engines available, powering production AI systems that demand traceable, hallucination-free answers from complex enterprise data. The platform processes PDF, DOCX, Excel, and PPT files through vision-based deep document understanding with layout analysis and OCR, extracting structured knowledge from tables, charts, and images that simpler parsers miss entirely. RAGFlow's hybrid retrieval pipeline combines vector search with BM25 keyword matching and multi-stage reranking across configurable document stores including Elasticsearch, InfiniFlow's Infinity engine, OpenSearch, and OceanBase. Developers connect any combination of LLM providers — OpenAI, DeepSeek, Anthropic Claude, Google Gemini, and locally-hosted models via Ollama — through a unified configuration layer. The visual agent workflow system enables multi-step reasoning chains with persistent memory, tool calling, and pre-built templates for common enterprise scenarios. RAGFlow synchronizes data from Confluence, S3, Notion, and Google Drive, and delivers answers through chat integrations with Feishu, Discord, Telegram, and Line. The Python SDK and RESTful API on port 9380 provide programmatic access to knowledge base management, document parsing, and conversational retrieval. The full stack deploys via Docker Compose with MySQL for metadata, Redis for task orchestration, and MinIO for object storage. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache 2.0 licensed.

Chat with your own documents: AnythingLLM, from Mintplex Labs, wraps retrieval-augmented generation (RAG) in an open-source application anyone can run. You organize content into workspaces, each an isolated namespace with its own documents, vector embeddings, chat history, and settings, so one instance can hold several separate knowledge bases. Upload PDFs, DOCX, TXT, and other formats, or scrape web pages; the built-in collector parses and chunks them into a vector database (LanceDB by default, with Pinecone, Chroma, Qdrant, and others supported). Answers cite their source documents. It works with both cloud LLMs (OpenAI, Anthropic, Gemini) and local ones via Ollama or LM Studio, and the embedding model is separately configurable. Beyond RAG chat, it includes AI agents that can browse the web and run tools, an embeddable chat widget for your website, a developer API, and multi-user mode with admin, manager, and default roles plus per-workspace access control. Context assembly is smarter than naive RAG: pinned documents, attached files, vector search hits, and recent chat history are combined under a token budget so the model's context window is filled efficiently, and each workspace supports multiple independent conversation threads against the same knowledge base. Because the embedding model, vector store, and chat LLM are all independently swappable, you can move between providers without re-ingesting a single document. The stack is Node.js with a React frontend, MIT-licensed.

Kotaemon is a document QA platform that combines advanced RAG techniques with a clean Gradio-based web interface for chatting with your documents. Built by Cinnamon, the Python backend supports any LLM provider including OpenAI, Azure OpenAI, Cohere, Groq, and local models via Ollama and llama-cpp-python, with a model management panel for configuring LLM and embedding providers from the UI. The default hybrid RAG pipeline combines full-text keyword retrieval with vector similarity search and applies re-ranking to ensure optimal result quality, while multi-modal document parsing extracts content from tables and figures alongside text. Advanced citations link every answer to specific source passages with relevance scores, viewable directly in the built-in PDF viewer with highlighted text spans. GraphRAG indexing via NanoGraphRAG, LightRAG, or Microsoft GraphRAG builds knowledge graphs from document collections for relationship-aware retrieval. Agent-based reasoning supports question decomposition for multi-hop queries using ReAct and ReWOO strategies. Multi-user authentication organizes documents into private and public collections with sharing and collaboration features. The platform supports Docker deployment in lite, full, and Ollama-bundled variants, runs on port 7860, and stores application data in a persistent volume. MCP tool integration enables external system connections for extended retrieval capabilities. On RepoCloud, deploy Kotaemon on a dedicated VPS with Docker, root SSH access, and complete control over your document AI infrastructure, all under the Apache 2.0 license.

Plain chunk retrieval hands an LLM disconnected excerpts; LightRAG builds a knowledge graph from your documents so answers can join facts spread across many files. Ask about one named person, product, or clause and local mode pulls its details. Ask for themes across the whole library and global mode follows the links between entities. Hybrid blends the two, naive falls back to plain chunk search, and mix combines all three. Mix is the default. Upload text, Markdown, PDF, Word, PowerPoint, and Excel files; MinerU, Docling, or the native parser can extract images, tables, and formulas. The admin console manages documents, explores the graph visually, and debugs queries, while a query-only workspace page gives everyday colleagues a clean chat screen that works on phones. A REST API covers indexing and queries, and an Ollama-compatible endpoint lets chat front ends such as Open WebUI use the knowledge base as a model. New files merge into the existing graph without a full reindex, and deleting a document rebuilds only the entities and relationships it touched from cached results. Separate model roles let an inexpensive model extract entities while a stronger one writes answers. An API key or accounts protect the server, and named workspaces keep several knowledge bases apart. Pair it with local Ollama models and no document text leaves the server. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

When a knowledge base gives a wrong answer, MimirQ shows which stage let it down. Documents move through a pipeline that teams can open at every step: parsing, cleaning, chunking, indexing, retrieval, reranking, and cited answers. A dataset catalog tracks documents, chunks, and ingestion status, and an execution monitor charts throughput, retries, parser queues, and run logs. The governance workbench previews each file, scores quality, and supports cleaning and labeling before chunks are cut by headings, records, or parent-child structure. Retrieval blends keyword and vector search with reranking, and every reply lists its source evidence with feedback scoring. A graph view maps entities, events, and relationships on one canvas for exploration. Golden-set evaluation stores standard questions and answers, runs them against the current pipeline, and reports recall and ranking metrics, so a release can be blocked when quality slips. Settings pages manage model providers, parsers, and retrieval options, and a Dify-compatible external knowledge API lets existing workflows call it. Role-based access and document permissions keep restricted material hidden from the wrong users. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Turn scattered meeting notes, policies, postmortems, and documents into a readable wiki that your team can question in plain language. Paste text, upload a file, or point KaaS at a web page, and an AI pipeline pulls out the concepts, decisions, and people, then writes them up as Markdown articles sorted into decision, reference, guide, and concept folders. Ask a question in chat and the answer cites the exact articles it came from, so anyone can open the source and check it. Each article shows tags, a source count, and a table of contents, and every page is stored as a plain Markdown file you can read, edit by hand, or commit to version control. Adding one new document only recompiles that document, which keeps model costs predictable as the library grows. A Builds page lists every submission with its status and attempts, and a failed worker's job is picked up again instead of lost. Derive a topic knowledge base to carve a focused collection out of the wiki for one subject, and let coding agents query the same wiki through an MCP ask tool. Team handbooks, customer onboarding, incident history, and pricing rules are the knowledge that ends up scattered across a dozen files, and this is where it comes back together. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Your documents change every day, and your RAG answers should change with them: Serviette keeps a chat and REST API continuously in sync with the files you point it at, no re-indexing jobs to remember. Point it at a local folder, Google Drive, S3 or MinIO bucket, SharePoint library, or even an FTP or WebDAV share, and edits, additions, and deletions flow into the index in real time. One YAML file, or the interactive wizard, chooses among eight vector databases, from embedded DuckDB with zero setup to pgvector, Qdrant, Milvus, Chroma, Weaviate, Pinecone, or MongoDB Atlas. Text, Office files, PDFs with tables, and scanned images are parsed and answerable by default, with audio and video joining when you add the keys. Embeddings can run fully locally with no credentials, and keyless answers simply quote the retrieved passages with their sources. The indexer and the stateless server scale independently, the server shards across workers with one config line, and both a ChatGPT-style web chat and a versioned /api/v1 retrieval API answer on the same port. A free license key from Pathway activates the engine, and benchmarks cover corpora up to 50 GB of text. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Stop pasting your files into ad-riddled online converters and handing strangers copies of your documents: HRConvert2 runs the whole conversion workshop on your own server instead. Drop a file onto the browser page and pick an output format from a catalogue of 488, covering documents, spreadsheets, presentations, ebooks, comics, images, audio, video, subtitles, and archives. Built-in OCR reads scanned PDFs and images and hands back editable text documents you can search and reuse. It converts 3D models and CAD drawings, renders OpenSCAD projects with their multi-file assemblies resolved, and can even build bootable ISO disk images for x86 and ARM machines from ordinary uploads. Live streams are captured and re-encoded straight from .m3u8 playlists, and every uploaded file can be virus-scanned with ClamAV before or after conversion. Finished files go back out through temporary share links that expire on their own, so sending a converted video to a client takes one click. Each visitor works in an isolated scratch space, files delete themselves when they age out, and nobody needs an account, ever. When traffic spikes, the server watches its own CPU and memory, throttles conversions, and spreads storage across disks so it stays upright. Three interfaces, seven color schemes, and 26 languages let each user shape the page to taste, and the whole thing runs happily on a Raspberry Pi with no database at all. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPL-3.0 licensed.