Research by UNMS
Research is a simple, beautiful desktop app for reading and thinking through research papers. You can upload PDFs, highlight, take notes, and use AI to summarize and connect ideas, built for anyone who wants to actually engage with what they read.
Get to know Research by UNMS
Who’s behind Research?
Hey, I’m Udara – maker of Research, Pile and Distill.
I've always been into a bunch of things. Growing up, I loved computers and creative work – taught myself design and code through online communities. I also loved science, so I studied bioinformatics in university, which let me blend both interests.
After graduating, I launched Clew, which built collaborative tooling for digital work. I worked on it for a few years before we were acquired by Dropbox, where we worked on incubating new products. After that, I wanted to get back to making things of my own, so I took a break and started building a few tools I wanted to use at the time.
That's how Research happened. I wasn't trying to build a company – I just wanted a better way to read research papers. Something simple, beautiful, and worth coming back to.
What’s Research and what’s so cool about it?
Research is a clean, focused desktop app for reading and thinking through research papers.
If you’ve ever tried to build a habit around reading technical PDFs, you probably know the pain: they’re dense, isolating, and hard to come back to. Tools like Zotero are great for reference management, but not for actually engaging with what you read.
Research is different. You drop in a paper, and it gives you space to read, annotate, highlight, but also to ask questions, reflect, and even chat with the content. The AI helps where it makes sense: summarizing, surfacing related entries, guiding you back to ideas you’ve touched before. It’s the kind of tool that quietly supports you without getting in the way.
The app is built for people like me: people who read a lot, think across topics, and need a place to connect ideas. It’s designed to feel like a quiet, earthy workspace. You can open multiple windows, each paper color-coded for quick context. The palette is soft, the layout is simple, and everything is focused on keeping you in flow.
I picked the colors intentionally – soft, earthy tones that make it easy to glance between multiple papers without getting tired. Most PDFs are stark white, and I didn’t want to stare at that for hours. So Research adapts the background to the paper, keeping things easy on the eyes and more natural to read.
Since launching, it’s grown fast, getting over 1,000 downloads a day. Early on, it was invite-only – I kept it in beta while I gathered feedback from friends and a small group of users. Once I opened public downloads, it really took off. The response has been way beyond what I expected.
And while it’s still early, the feedback has been incredible. It’s not just for academics. People in biotech, aerospace, AI. Even lawyers and product teams are using it to read complex documentation. It fills a gap between the chaos of browser tabs and the rigidity of citation managers. And that’s exactly what I hoped it would do.
Tool Stack of Research by UNMS
What’s under Research’s hood? Which technologies were used and why did you chose them?
Research is built with Tauri. I moved away from Electron for this one. I love Tauri’s plugin system and the way it integrates with Rust. It’s cross-platform, lightweight, and surprisingly ergonomic once you get used to the v2 documentation. And it doesn’t bundle Chromium, which helps keep the app small and responsive.
On the frontend, I use React. For the backend, I run Laravel and Rails – I bounce between the two depending on the project. I’ve always been drawn to their structure, readability, and developer experience. Laravel, in particular, is something I’ve been using since I was first learning to build full-stack apps.
For LLM features, I mostly use OpenAI models: their tooling is solid, and they cover most use cases well. For longer documents or research-heavy tasks, I sometimes switch to Gemini, since it handles larger context windows better. I’ve also been experimenting with local models like DeepSeek for more self-contained workflows.
I also use AI as part of my development process – not to write entire features, but to move faster on small parts. It lowers the cost of experimenting and makes it easier to try new ideas.
Everything’s hosted on my own servers. I don’t use serverless. Not because I’m against it, but because I prefer having full control. A simple $5 DigitalOcean droplet can go a long way when you’re careful with your stack.
Do you use any other tools to run the business?
I try to keep things simple. Most of my design happens directly in code, I don’t use Figma much anymore. I used to design in Photoshop as a kid, so over time I built up the muscle memory to sketch UI straight into a code editor.
The websites are mostly single pages, built to describe, not pitch, the product. I let the product speak for itself. I’ve always preferred to show rather than tell. There’s so much software out there that promises more than it delivers. I’d rather be upfront and let the product speak for itself.
For feedback and lightweight analytics, I use my own internal tool called Signal. You’ll see it embedded inside Research, it helps me collect direct user feedback and summarizes it for me using AI.
I also use Obsidian to archive ideas and notes, and Transmit by Panic to manage files on my servers. It’s one of those rare tools I’ve kept using for years.
What’s your personal stack? Which apps do you and your team love?
Honestly, my personal stack is minimal. I use Gmail in the browser, Chrome for browsing, and Obsidian for long-form notes. I don’t use a to-do app. I use paper notebooks. Every day gets a new page. I fill it with tasks, thoughts, sketches, and just flip to the next when the day’s done. It works. And it helps me stay close to the work.
I also still journal the old-school way. I’ve been writing for over a decade, dozens of physical journals stacked up over the years. That’s where a lot of the inspiration for Research came from.
Most of the tools I love are the ones that don’t try too hard. That stay out of the way. That feel more like instruments than apps.
Anything else you’d like to share?
I’m working on something new, a kind of journal-meets-agent experience called Distill. It builds on what I learned from Pile and Research, but with a more proactive AI layer. Think: a private space where you’re surrounded by your thoughts and a few configurable AIs that help move things forward: remembering tasks, surfacing ideas, following up.
It’s still early, but the initial response has been amazing, and I’m excited about where it’s heading.
In the meantime, if you spend your days reading research papers, long docs, or dense materials, give Research a try. It might be just the companion you’ve been missing.
Now, discover Research by UNMS for yourself
Huge thanks to Udara for sharing the story behind Research and the details on the building blocks that make it such a unique research tool. Go give it a spin, and while you’re at it, explore his other beautiful apps, Distill and Pile. Enjoy!