PandaWiki
In seven wizard steps, PandaWiki sets up a product docs, manual, FAQ, or blog site whose built-in AI answers visitors using only your own content. Those visitors can chat with an assistant that cites its sources and accepts image questions, or switch to a docs-only search that lists ranked pages with summaries. Publishing stays under control: each document has three switches, visible in navigation, reachable by URL, and answerable by AI, and nothing appears or gets learned until it is published. Authors write in a Markdown and HTML editor, or import from URLs, RSS and Atom feeds, sitemaps, offline .txt, .md, .xlsx, .docx, and .pdf files, Notion, Yuque, SiYuan, Wiki.js, Feishu, Confluence, and DingTalk. Pages export to Word, PDF, or Markdown. The wizard covers models, domain, HTTP or HTTPS ports, import, publishing, answer testing, and page decoration. One knowledge base can also feed a web widget and DingTalk, Feishu, WeCom, and Discord bots. Administrators get a dashboard of visits, questions, visitor regions, hot documents, and referring domains. Chat models come from Baizhi Cloud, DeepSeek, OpenAI, or Ollama, while vector and reranking models are built in. The open-source edition covers one wiki site, 300 documents, and one admin, and its console and documentation are in Chinese. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.
AnythingLLM
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.
Open WebUI
Large language models get a polished front end that can run fully offline: Open WebUI is the self-hosted front end of choice. It talks to local model runners, primarily Ollama, and to any OpenAI-compatible API, so LM Studio, vLLM, Groq, Mistral, OpenRouter, and cloud providers all plug into the same chat interface and can be mixed per conversation. RAG is built in: upload files to knowledge bases or reference them in chat with the # command, backed by a choice of nine vector databases (ChromaDB and PGVector officially maintained) and multiple extraction engines including Tika and Docling, with hybrid BM25-plus-vector search and cross-encoder reranking. Web search results from providers like SearXNG, Brave, and Tavily inject directly into conversations. Extensibility comes from Python tools and functions that run inside the chat, a Pipelines plugin framework, and native MCP support. Multi-user features include RBAC, SSO, and group permissions, and the instance itself exposes an OpenAI-compatible API your own apps can call.
Heym
Heym lets you build AI agents and workflows on a visual canvas, run them from APIs, schedules, or chat, and watch every execution, token, and cost. Describe a system in text or voice and the AI Assistant draws the graph. Prefer hands-on? Compose it from LLM and Agent nodes, conditions, switches, loops, merges, and sub-workflows. An orchestrator calls named sub-agents in parallel. RAG runs on Qdrant or pgvector, and a knowledge graph keeps facts across runs. Triggers: cron, webhooks, messaging integrations, inbox events, WebSocket messages. Playwright nodes automate browsers, and Codex or OpenCode steps work on repositories. Checkpoints pause an agent so a person can accept, edit, or refuse before the run resumes. Every run lands in History. Traces list model requests, tool calls, tokens, latency, and USD cost, and alerts fire on errors, duration, spend, or run count. Evals compare prompts across models. OpenTelemetry exports reach Jaeger, Grafana Tempo, or Datadog. Publish workflows as REST or SSE endpoints, as MCP tools for Claude Desktop and Cursor, or as public Portal chat pages, then add dashboard widgets, Data Tables, a Kanban Board, and Drive file sharing. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Source-available under the MIT License with the Commons Clause.
Lumen
Ask Lumen which islands have the most penguins and it writes the SQL, runs it, and draws the chart in a split-screen explorer. Drag in CSV, Parquet, JSON, or Excel files, or point it at a database, and mark extra files as metadata so the model understands what each column means. Every answer lands in its own exploration, with follow-up questions nested beneath it, and the generated SQL or Vega-Lite chart specification sits in an editor you can change to update the view instantly. Tables sort, filter, and paginate, charts carry tooltips and annotations, and results export as CSV, Excel, PNG, or PDF. Report mode gathers all explorations into one document that downloads as a notebook with every question, query, and visualization ready to rerun. A planner agent breaks hard questions into steps, a validation pass checks that the answer matches the request, and settings choose among OpenAI, Anthropic, Ollama, and other language model providers. Developers can add custom agents, tools, and analyses, or write declarative specifications that become shareable dashboards. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. BSD-3-Clause licensed.
RocketRide
Building an AI pipeline in RocketRide means dragging nodes for models, vector stores, document parsers, and agents onto one visual canvas and wiring their typed lanes together, with no boilerplate in between. Press play on a webhook, chat, or file-dropper source node to start the pipeline. The catalog holds 170 nodes across 21 types, including 16 LLM nodes for OpenAI, Anthropic, Gemini, Mistral, and Ollama, vector stores such as Qdrant, Pinecone, Milvus, and pgvector, OCR, entity recognition, PII anonymization, reranking, and database connectors. Agent nodes run CrewAI, LangChain, LlamaIndex, or Deep Agents, and a persistent memory node carries context from one run to the next. Every pipeline is plain JSON. Commit it, diff it, review it, then run the identical file on a laptop, a server, or RocketRide Cloud. Tracing shows call trees, token usage, latency, and memory for each run, so you can compare models, agents, and tools before scaling up. Applications call pipelines through TypeScript and Python SDKs, and coding agents such as Claude and Cursor can build, change, and deploy them in plain language. Example pipelines cover document intake that reads text from scans and masks personal data, question answering over private files, and agent workflows, while Discord and Telegram source nodes feed chat bots into the pipeline. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
OpenAgent
Chat, research, and everyday chores all run through OpenAgent, a private AI assistant hosted on a server you control. Pick from more than thirty model providers, including OpenAI, Anthropic, Gemini, DeepSeek, and local Ollama models, and switch models inside a conversation. Agents can drive a real browser, search the web, run shell commands, read and write Word, Excel, and PowerPoint files, and call any MCP server. Every tool call shows its arguments and result. Upload PDFs and spreadsheets to build isolated knowledge bases with semantic search, then attach them to a chat. A drag and drop workflow builder chains steps with branches, parallel tasks, and schedules. The admin console tracks token use and cost per user and model with charts and a message heatmap, lists every tool, records audit logs, and shows live CPU, memory, and API latency. Single sign-on through OIDC, OAuth2, LDAP, or SAML and separate workspaces keep teams apart, and a REST API with Swagger docs exposes every feature. 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.
ClickHouse Loom
Give every AI agent you use the same memory, so what Claude Code learns on Monday is waiting for Codex or Cursor on Tuesday. Loom is a self-hosted server that stores memories, documents, conversations, and a code index, then serves them to agents over MCP, HTTP, and Python or TypeScript SDKs. Each fact is saved under a key with history and provenance, near-duplicates merge, and contradictions land in a review queue where a person keeps one side, drops one, or allows both. Search can abstain instead of guessing, and the Ask AI panel answers with numbered citations back to the memory or document behind each claim. A library indexes PDF, docx, and pptx files, and the repository explorer finds symbols, callers, and change impact with the commit it answers from. Namespaces, roles, and tokens let teams share memory safely, an audit log records every save, delete, and grant, and erasure requests are verified. Task inboxes pass work between agents with leases and retries. One installer script starts Loom and ClickHouse, and plugins connect Claude Code and Codex. 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.
LangBot
Connect one AI assistant to every chat app your customers already use, then manage it from a single admin panel. Create a bot for Slack, Discord, Telegram, LINE, Lark, DingTalk, WeCom, Matrix, KOOK, or QQ, or embed the Page Bot widget on a website with one script tag. Event routing sends each message to the processor you choose, and a pipeline controls triggers, safety rules, the model and prompt, and how replies are sent. A built-in Debug Chat streams answers as you edit, in private or group sessions. Knowledge bases split uploaded documents into chunks and search them by vector, full-text, or hybrid matching, with query rewriting and reranking options. Add providers like OpenAI, Anthropic, DeepSeek, Gemini, and Ollama, then give each pipeline its own models. The dashboard counts executions, success rate, token usage, and active sessions, and tabs list model calls, user feedback, and error logs. Sessions show the full conversation for each user, with a button to reset context. Install plugins and MCP servers from the extension marketplace to add tools. Use it for a support bot, a community assistant, an internal helpdesk, or a multi-channel front desk. 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.
FastGPT
FastGPT lets you build production AI agents and knowledge base chatbots through a visual drag-and-drop workflow editor, connecting any LLM provider to your documents with retrieval-augmented generation that cites sources and reduces hallucination. The workflow canvas chains LLM calls, conditional branching, HTTP requests, code sandbox execution, and plugin nodes into complex conversation flows and agent skill pipelines without writing backend code. The knowledge base engine ingests documents in ten formats (TXT, Markdown, HTML, PDF, DOCX, PPTX, CSV, XLSX, URL scraping, and CSV batch import) then applies automatic chunking, hybrid vector retrieval with semantic reranking, and QA-pair splitting to deliver accurate, citation-backed answers. FastGPT connects to virtually any LLM provider through its AI Proxy aggregation layer: OpenAI GPT-4o, Anthropic Claude, Google Gemini, DeepSeek, Qwen, ERNIE Bot, and models hosted via Ollama all work through a unified OpenAI-compatible API. Bidirectional MCP support enables agents to call external tools and expose their own capabilities to other systems. Completed applications can be shared via login-free links, embedded as iframe widgets, or integrated with WeCom, Lark, DingTalk, and WeChat Official Accounts through the published REST API. Application operation logs, conversation annotation, and per-model usage analytics provide full lifecycle governance for compliance-sensitive deployments. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. FastGPT Open Source License (Apache 2.0 based) licensed.
EdgeQuake
A folder of documents stops being a pile of files once EdgeQuake extracts a knowledge graph you can explore and question. Upload text, PDFs, or markdown and the pipeline extracts people, organizations, places, and the relationships between them, then shows progress and entity counts for every file. The graph viewer draws the network with color-coded entity types, a searchable entity list, and a details panel that lists descriptions, properties, and linked entities. Ask a question in the query chat and pick a retrieval mode, from keyword lookups to hybrid and global themes, and each answer arrives with numbered sources and a button that highlights the matching nodes on the graph. Workspaces keep projects separate, and each can assign a different model to extraction, answering, keywords, summaries, and embeddings. A cost dashboard breaks token use and spend down by stage with a monthly budget bar. A members page, an API explorer, and Swagger docs round out administration. 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.
Open Notebook
The most feature-complete open-source alternative to Google's NotebookLM — a self-hosted research platform where you upload PDFs, videos, audio files, and web pages into organized notebooks, then chat with your content, generate multi-speaker podcasts, and run semantic search across everything without sending a single byte to Google's servers. The podcast engine supports 1-4 fully customizable speakers with backstories, personalities, and expertise profiles, generating professional audio dialogue through OpenAI, ElevenLabs, Google TTS, or completely local text-to-speech via Kokoro for maximum privacy. Content processing uses token-based chunking with RAG-powered retrieval grounded in your uploaded sources, while both full-text keyword search and semantic vector search via SurrealDB enable conceptual discovery across all notebooks. The 18+ supported AI providers include OpenAI, Anthropic, Google Gemini, Groq, Ollama, LM Studio, and more — configurable per task so you can route cheap models to summarization and powerful models to analysis. Content transformations extract insights, generate summaries, create study guides, and produce structured outputs from any source material. The MCP integration connects Open Notebook to Claude Desktop, VS Code, and other MCP clients for seamless workflow integration. A full REST API on port 5055 enables complete automation of notebook management, source upload, and podcast generation. Deploy via Docker Compose with the application container, SurrealDB v2 on RocksDB, and optional TTS containers. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
BullFlow
Generic AI writing sounds like nobody in particular; BullFlow writes from your own company knowledge instead. It first builds a private knowledge base from your documents, images, audio, and video plus structured records for your company, brand, products, customers, and cases. Ingestion runs entirely on local open models: OCR reads scans, semantic splitting and chunking prepare text, and hybrid vector plus full-text search retrieves passages with their sources and relevance scores shown under every answer. A creation agent then works in one conversation: it searches the knowledge base, drafts articles with images embedded in the text, generates image sets or short videos, and exports finished pieces as PDF or Word. Output adapts to where it is going, long-form articles for article platforms, multi-image posts with matching copy for image feeds. A separate distribution gateway sends one finished piece to many accounts across Chinese and international social platforms, checks account health on a schedule, and returns publish results and metrics by webhook. Around all of that sits a commercial shell: multiple tenants with isolated data, roles from super admin to finance, credit billing measured per token, image, and second, subscriptions, invoices, referral commissions, and a conversion funnel in the operations backend. Scheduled rules can generate and publish on a timetable with nobody watching, and notifications report each completion or failure. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
GEOFlow
Built for brands that want AI search engines to quote them, GEOFlow is an open-source admin system that turns approved knowledge into reviewed articles and publishes them across every site you run. Load documents or import trusted web pages into knowledge bases that are chunked into evidence units, with optional semantic planning and vector retrieval. Combine them with title libraries (AI can draft up to 100,000 titles), keywords, images, authors, and prompts, then set a task's generation limits, cadence, and publishing scope. Chat and embedding models connect through OpenAI-compatible endpoints or native Gemini, with automatic fallback to the next model when a call fails. Each draft passes an AI quality gate that checks evidence, citations, advertising rules, and publishing context, records scores with source locations and revision advice, and holds risky articles for a human reviewer. Approved articles appear on the built-in site with SEO metadata, Open Graph, Schema, sitemaps, and llms.txt, or travel to hosted subdomain sites, GEOFlow Agent packages, WordPress through its REST API, or any HTTP endpoint. A manual publishing workbench and Chrome assistant cover platforms that need a person, while analytics track traffic, AI crawlers, and brand visibility in generative engines. An API, command-line tool, and agent skill support automation. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.
ByteChef
Drag an AI agent onto a canvas beside a Gmail trigger, a condition, and a Google Sheets step, and ByteChef runs the whole thing as one durable workflow that resumes where it stopped. Runs that fail or wait for a human pick up from the step they reached. The visual editor maps values between steps with data pills, and its 250+ connectors double as agent tools and MCP tools. An agent combines a model from twelve LLM providers, including OpenAI, Anthropic, Mistral, and Ollama, with chat memory, knowledge-base retrieval, and tools. It streams structured answers. Flow controls cover conditions, branches, loops, parallel and fork-join paths, subflows, error handling, and approval steps that pause for a person. Workflows start from webhooks, polling, schedules, forms, or app events, and code steps run JavaScript, Python, or Ruby. Workflows can be exposed through an MCP server for Claude Desktop, Cursor, or Windsurf, and ByteChef consumes external MCP servers too. Data Tables add typed relational tables with a spreadsheet-like grid and CSV import and export. The Agent Playbook opens a chat with the agent as you build it. Every run is listed with per-step input, output, and errors. 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 + EE licensed.
Serviette
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.
BISHENG
Document review, fixed-layout report generation, and multi-step research that ends in a PPT, Excel, or Word file all run inside BISHENG, a self-hosted workspace for enterprise AI applications. Review and report flows are drawn on a visual canvas where a circle becomes a loop, side-by-side nodes run in parallel, and a multi-item selection runs as a batch. A running flow can stop for a person to give input, confirm, or make changes, and chat and automation share one flow type. Research goes to Task Mode: give it a goal plus any files, knowledge bases, tools, and skills, and it writes a checklist, hands research to parallel sub-agents, and returns a report with clickable citation markers. Teams sharing documents use Knowledge Spaces, with folders, tags, keyword search, AI Q&A that shows its sources, and permissions down to a single file for people, departments, or user groups. For a daily news briefing, Channels turn WeChat official accounts and websites into topic feeds and save articles as knowledge in one click. Admins connect OpenAI-compatible APIs, Azure OpenAI, Anthropic, Qwen, DeepSeek, and local Ollama models, then govern access with approvals, audit logs, SSO, and LDAP. Private document parsing reads printed text, handwriting, tables, layouts, and seals, and a service-account Open API lets other systems call applications and knowledge bases. 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.
LightRAG
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.