I paired a local LLM with Obsidian on my phone, and it's the productivity boost I didn't expect

Published Aug 1, 2026, 7:30 AM EDT Nolen began their writing career in 2019, with three years dedicated to editing the Creative section at MakeUseOf. Their expertise lies at the crossroads of technology and creativity, covering areas like photography, video editing, and graphic design. Outside of work, you'll often find Nolen diving into a good book, writing their own stories, or playing video games. Running a local LLM on desktop is one thing. Pairing it with an Obsidian vault is where the setup actually starts becoming more useful for a daily workflow. And this can be easily done with either an MCP setup or a filesystem connector. On mobile, though, that whole workflow just doesn't exist. There are AI plugins in the Obsidian community that support local models, but not one of them runs the model on the actual phone. Not only that, but my default mobile local LLM app also couldn't accept documents. So it wasn't until I started digging around for more feature-rich local LLM apps that I was finally able to get a real Obsidian-to-AI pipeline going on my phone… Want to stay in the loop with the latest in AI? The XDA AI Insider newsletter drops weekly with deep dives, tool recommendations, and hands-on coverage you won't find anywhere else on the site. Subscribe by modifying your newsletter preferences! Mobile Obsidian is a filing cabinet you can't really search My vault lives on my phone, but I can barely use it there My Obsidian vault sits in local phone storage, not iCloud. I've had it that way in order to keep it genuinely on the device instead of on Apple's servers. Regardless of where your vault lives, though, the trade-off nobody really talks about is that mobile Obsidian is great for capture and pretty rough for retrieval. Dropping a quick note in is fast. Actually finding something a week later, when I don't remember which folder I put it in or exactly what I called it, is where it starts to put a spanner in things. The obvious answer to this is an AI plugin, and Obsidian's community has plenty of them. On desktop they're honestly great, especially wired up to a local LLM through Ollama or LM Studio. But on mobile, even the ones that support local models can't actually run those models on the phone. They reach back to your desktop over LAN or Tailscale, which technically works but isn't really local anymore, it's just remote with extra steps. So even though I've built a desktop workflow exactly like this, it's useless to me when I'm away from my desk. All I needed was a mobile local LLM app that could read documents An app that finally lets me read my own notes I've been using PocketPal for a while as my mobile local LLM tool of choice. It runs models on-device and has decent UI/UX, so it's been my go-to for quick offline chat. However, like most other mobile runners, it doesn't accept document upload. It's pretty rare for local LLM tools on mobile to have RAG systems because of the hardware restrictions. So I did some digging and came across Noema, a local LLM runner that does actually do document ingestion on the phone itself, and it was the thing that made me switch over. It is an iOS-only app at the time of writing, so if you're on Android, I recommend checking out LocalRAG. Noema is free, and runs GGUF and MLX and Liquid AI's SLM formats natively, which is already more flexible than PocketPal offers. But the reason I chose it isn't for the model support, it's what happens when you attach a document. The first time you try to add a file, Noema downloads an embedding model which is what turns your text into vectors it can actually search. It's a one-time setup, and after that every document you add gets indexed on-device with no cloud calls at any point. Once that's done, you just hit the Plus icon, find your vault in the Files app, and add your .md and .txt files as documents. It handles Markdown natively, along with PDF, EPUB, TXT, CSV, JSON, and JSONL, so all my .md notes are fair game, and the PDFs I've dropped into my vault work too. I recommend keeping an eye on your context window, though, because documents can eat into it fast. What separates this from generic "chat with your files" is that Noema cites the note it pulled the answer from. Ask it about something you've written about across a few notes, and it comes back with a synthesized response with numbered references to the actual chunks it pulled. That's retrieval, not summarization, which is exactly what I wanted on mobile. Regardless of what OS you're on, I recommend going for a tool that accepts documents from your files app and has some type of RAG system built-in. That's your best shot at getting an offline, AI-first Obsidian workflow. The whole point of this setup is that it works when nothing else does No signal, no problem Why even bother with this when I can just use Claude on mobile? Well, I don't have a connection at all times. Sometimes the Wi-Fi cuts out, sometimes the power goes out, sometimes the cell towers go down, and other times I'm traveling. But now I still get full responses with cited notes, the same way I would have with a cloud-based chatbot. Except the cloud AI doesn't die the second I lose signal or turn on airplane mode. I'd also argue that Noema is more feature-rich than apps like Gemini and ChatGPT. I can bookmark a response, pin a response to the scratchpad (Noema's notepad window), branch it, or just copy it with formatting intact straight into Obsidian. Of course, a local LLM runner also exposes controls like temperature and context length, giving me complete customization over the workflow. So overall, it's just the better tool to integrate with something like Obsidian - they have the same philosophy of ownership and control. The mobile AI story has been dominated by cloud tools for so long that the idea of running the whole stack on your phone - model, embeddings, and your own knowledge base - still feels like it shouldn't really work. Then you try it once with no signal and realize it does. If you've got a mobile vault and a few gigs to spare, this is definitely a setup worth trying out. Obsidian OS Windows, macOS, Linux, iOS, iPadOS, Android Individual pricing Free normally; $4/month for Obsidian Sync