The latest articles on DEV Community by Sadot Hdz (@zademy). https://dev.to/zademy https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3921103%2F4137ce02-0bb0-4f17-ae61-decf63aa3bc5.jpeg https://dev.to/zademy en Sadot Hdz Sat, 09 May 2026 05:18:10 +0000 https://dev.to/zademy/i-built-an-open-source-ai-workspace-for-paperless-ngx-because-i-wanted-document-intelligence-mdf https://dev.to/zademy/i-built-an-open-source-ai-workspace-for-paperless-ngx-because-i-wanted-document-intelligence-mdf <h1> I Built an Open Source AI Workspace for Paperless-ngx Because I Wanted Document Intelligence </h1> <p>A few months ago, I started running into a problem that many developers and small teams probably already know:</p> <p>We store a huge amount of documents everywhere.</p> <p>PDFs. Notes. Contracts. Scanned files. Technical documentation. Personal archives. Random folders named <code>final-final-v2</code>.</p> <p>And even after organizing everything with tools like <a href="https://docs.paperless-ngx.com/" rel="noopener noreferrer">Paperless-ngx</a>, I still felt like something was missing.</p> <p>I didn’t just want document storage.</p> <p>I wanted document intelligence.</p> <p>So I started building <a href="https://taan-mind.com/" rel="noopener noreferrer">Taan Mind</a>.</p> <h2> The idea </h2> <p>The main goal behind Taan Mind was simple:</p> <blockquote> <p>What if your document archive could actually understand your files?</p> </blockquote> <p>Instead of manually searching through folders or reading entire PDFs, I wanted a workspace where AI could:</p> <ul> <li>understand documents</li> <li>extract metadata</li> <li>enrich OCR content</li> <li>connect conversations with files</li> <li>work with local-first AI models</li> <li>stay self-hosted and privacy-focused</li> </ul> <p>The result became an open-source AI workspace built around Paperless-ngx.</p> <h2> Why I built it </h2> <p>Most AI document tools today have one big issue:</p> <p>They require uploading sensitive files to external cloud services.</p> <p>For many people that’s fine.</p> <p>But for developers, self-hosters, researchers, or small companies handling private data, that’s not always acceptable.</p> <p>I wanted something that could work locally with tools like:</p> <ul> <li>Ollama</li> <li>local LLMs</li> <li>Docker</li> <li>self-hosted infrastructure</li> </ul> <p>while still feeling modern and easy to use.</p> <h2> What Taan Mind does </h2> <p>Currently, the project includes:</p> <ul> <li>AI-powered chat with document context</li> <li>OCR processing pipelines</li> <li>Metadata enrichment</li> <li>KPI dashboards</li> <li>Paperless-ngx integration</li> <li>Local-first model support with Ollama</li> <li>Multi-provider AI support</li> <li>Docker-ready deployment</li> </ul> <p>The frontend is built with:</p> <ul> <li>Nuxt 4</li> <li>Nuxt UI</li> <li>Tailwind CSS</li> <li>AI SDK</li> </ul> <p>And the stack also uses:</p> <ul> <li>Ollama</li> <li>MuPDF</li> <li>SQLite</li> <li>Drizzle ORM</li> <li>Docker Compose</li> </ul> <p>The idea is to create a practical AI workspace instead of just another chatbot UI.</p> <p>Website:<br> <a href="https://taan-mind.com/" rel="noopener noreferrer">https://taan-mind.com/</a></p> <h2> One of the hardest parts </h2> <p>One of the most difficult things was balancing:</p> <ul> <li>privacy</li> <li>performance</li> <li>OCR quality</li> <li>AI context injection</li> <li>local model execution</li> </ul> <p>especially when trying to keep the project simple enough for contributors to run locally.</p> <p>I also wanted the architecture to stay modular so future integrations like MCP servers, new AI providers, workflows, and automation can be added later without rewriting everything.</p> <h2> Why open source </h2> <p>I decided to make the project open source because I genuinely believe document AI should be transparent.</p> <p>People should be able to:</p> <ul> <li>inspect the code</li> <li>run it locally</li> <li>customize providers</li> <li>choose their own models</li> <li>own their data</li> </ul> <p>I also think the self-hosted AI ecosystem is becoming incredibly important right now.</p> <p>Projects like:</p> <ul> <li>Ollama</li> <li>Open WebUI</li> <li>Paperless-ngx</li> </ul> <p>have shown how strong community-driven tools can become.</p> <h2> What surprised me the most </h2> <p>The biggest surprise wasn’t the AI part.</p> <p>It was how many people are searching for better ways to interact with their own documents.</p> <p>Not just “chat with PDF” demos.</p> <p>But real workflows like:</p> <ul> <li>document organization</li> <li>searchable archives</li> <li>OCR pipelines</li> <li>metadata automation</li> <li>AI-assisted knowledge retrieval</li> </ul> <p>That space still feels very early, and there’s still a lot left to build.</p> <h2> What’s next </h2> <p>Some upcoming ideas include:</p> <ul> <li>better RAG pipelines</li> <li>MCP integrations</li> <li>workflow automation</li> <li>semantic search improvements</li> <li>more local model optimizations</li> <li>collaborative document workflows</li> </ul> <h2> Final thoughts </h2> <p>This is actually my first post on Dev.to, so I wanted to share something I’ve genuinely been excited to build.</p> <p>If you’re interested in:</p> <ul> <li>self-hosted AI</li> <li>open source</li> <li>document workflows</li> <li>local LLMs</li> <li>Paperless-ngx</li> <li>Nuxt</li> <li>AI tooling</li> </ul> <p>I’d genuinely love feedback from the community.</p> <h2> Links </h2> <p>Website:<br><br> <a href="https://taan-mind.com/" rel="noopener noreferrer">https://taan-mind.com/</a></p> <p>GitHub:<br><br> <a href="https://github.com/zademy/taan-mind" rel="noopener noreferrer">https://github.com/zademy/taan-mind</a></p> ai nuxt paperless typescript