Why I Built NoteCove: Local-First in 2026

In my career, I have spent years working on collaborative editing infrastructure. First at Google on Docs, Spreadsheets & Apps Script, then at Dropbox working on Dropbox Paper. I know exactly what it takes to run real-time collaborative editing at scale. I wanted something capable of multiplayer editing, if only multiplaying amongst my own devices as it’s not unusual for me to quickly go back and forth between laptop, desktop, and/or phone. But also the ability to do so between work and personal devices as my organization doesn’t divide neatly between work and personal, some things are kinda both.

That being the case, it would have to be usable where I work now as well as the future. Companies I work for tend to care deeply about these sorts of things as compliance regimes are a thing. If I spun up a service, it’d be subject to all sorts of compliance concerns and approval. But, if I use storage they’ve already approved, the story is much easier to tell. And especially in 2026, any company I work at will have some form of cloud storage; be it Google Drive, iCloud Drive, Dropbox, whatever.

And also, with modern concerns about privacy and AI training on user data, if I’m building for not only me, I wanted something different. I wanted control over where my notes live. And with all of that, I definitely didn’t want to run another server as I don’t want to have to be debugging the service whilst in my day to day, as multiplayer editing means servers, replication, monitoring, operational complexity, and substantial ongoing cost. Using people’s own cloud storage that they already trust avoids all of that.

The Immediate Problem

In my case, the origin story was simple: I wanted to use Apple Notes at work. Clean interface, works great on Mac and iPhone, did most of what I need. But at work, we can’t use iCloud. No iCloud means no sync between my work laptop and work phone, no less having the things I want accessible in both work and personal devices. My notes were stuck.

I looked at alternatives. Every option had problems.

Evernote? Will the company even exist tomorrow? They’ve been struggling for years. Product has been janky for a while.

Notion? Great product, but your data lives on their servers. When I last checked, they were still doing paragraph-level sync, not true real-time like Paper had. Not a deal breaker, but I’d run into it a few times. And I can’t use it at work anyway.

Obsidian? Closer to what I wanted with local-first principles, but it never fit me. Markdown is nice for some things, but I wanted rich editing without having to think in markdown. I wanted to just write.

Google Keep? Google Keep is awful. And Google Docs don’t give me everything I want.

The pattern was clear: either there’s no sync, or you sign up for a service and hope they stay in business, respect your privacy, and don’t do things with your data you’d rather they didn’t, if they even provided what I wanted at all.

With AI coding tools, writing something myself in a reasonable time made this something where I could “scratch my own itch” on a side project.

Why Local-First Makes Sense Now in 2026 of all places

Could I have built NoteCove back in 2015-2016 when I was at Dropbox? Probably not in any reasonable timeframe. Three things have changed:

AI-assisted coding is real. I built NoteCove primarily using Claude Code. What would have taken months of grinding through boilerplate now takes weeks. This was partly an experiment to understand agentic coding at scale — where it works, where it doesn’t, what the gaps are.
CRDT libraries matured. Back in 2018, CRDT existed (it was invented in 2011), but widely available CRDT implementations barely existed. Now we have well performing CRDT libraries like Yjs, Automerge, and rich text editors like Tiptap built on top of them. The primitives are there. The alternative would have been to write my own Operational Transform code which while I could, was way more effort than I wanted to put in.
Cloud storage APIs are ubiquitous. iCloud, Google Drive, Dropbox, OneDrive — everyone has these. The APIs are stable and well-documented. Why run my own storage infrastructure when users already have storage they trust?

Also, honestly, when I was at Dropbox I had Dropbox Paper. I didn’t need to build a note-taking tool — I had one. It didn’t have everything, but had a lot. But now, looking at the landscape in 2026, I think there’s a place for something like this.

What NoteCove Is

NoteCove gives you notes and tasks where you control where your data lives, while still getting reliable sync across devices. Think of it as a personal Atlassian Confluence and Jira if you like; but one you’d want to use.

Your data can live in iCloud, Google Drive, OneDrive, Dropbox, or if you don’t need sync, on your own drive. Not on my servers. Not in some startup’s database that might disappear tomorrow. But rather in storage you already have and already pay for. Since it just stores files on disk, really any sync mechanism can work: like rclone or unison sync, for desktop anyway. Mobile is always a bit different.

Sync happens through CRDTs — the same conflict-free replication technology that powers modern collaborative editing. Your devices sync through your cloud storage. Offline-first, works without internet, no servers required. It’s not all that unusual for me to be in a place where network connectivity doesn’t exist, or I can only have one device on it at a time, so offline sync has to work.

The editor is inspired by Dropbox Paper: rich formatting, clean interface, with embeds, comments, links, emojis, and all the stuff you need. Just write. In 2026, it’s weird for the editor to be oriented towards print of all things (I’m looking at you Google Docs). Honestly, it’s something that drives me nuts at work : some docs I’d have to fill out had all sorts of unnecessary fancy formatting that I’d have to spend too much time making my additions match. There’s no need for that. Admittedly, in the days of AI, I probably don’t have to deal with that anymore anyway.

Who It’s For

Obviously, I built NoteCove for myself first. I needed something that worked the way I wanted to work. I think it can work for quite a number of people, but I had some specific groups in mind:

Writers: My daughter is a novelist with an MFA. Writers have specific needs: scratch pads for timelines, character studies, research notes alongside their manuscript. Tools like Scrivener feel dated. Tools like Dabble charge monthly rent. Google Docs doesn’t give you the structure you need. In talking with my daughter, what she needed mostly overlapped with what I’d already planned. And there are things that Dabble may never add that for me shoulnd’t be a heavy lift to implement, but also won’t clutter the UI.
Small workgroups: While I originally wrote it for me, I realized that it would work for small work groups to share notes, tasks, and so forth with ease. If you’re using something like iCloud or Google Drive while connected to the network, sync happens in a few seconds.
Creatives and Academics: Another angle I’m thinking about: proving human authorship. In the age of AI, writers and academics and other creatives might need to show they actually wrote something over time, not just dumped it from ChatGPT. NoteCove tracks document history naturally as part of its CRDT sync. The data is there to prove “I typed this over three weeks” if you need it. That feature isn’t built yet, but the foundation and the data already needed to power it is. I might be able to have something where you could prove that you in particular typed in the text at least.
People who want to pair with AI Agents: NoteCove has a CLI specifically so humans and AI agents can collaborate naturally.Your AI assistant can create and update tasks and notes alongside you, all stored in your own cloud storage. And in particular: I’ve used NoteCove to help build NoteCove itself. I wrote about my workflow in [Narrowing The Cone Of Error](https://Narrowing The Cone Of Error){.chip}, and after that post, updated it to put it’s plans and questions, etc. into NoteCove notes, and use NoteCove tasks for tracking. But I want to enable people to use agents with NoteCove, but not feel forced to.And when NoteCove grows better AI interactions, you’ll get to use your model and agent that you already know and pay for (or if you’re using ollama or lmstudio, etc. you don’t pay for).
People who need a solid organziation tool: Like I said before, it started because that’s what I needed, and what I had didn’t cut it.

Why Spend Nights, Weekends, and Vacation on This?

I’m a Staff IC at Datadog, I could spend my free time doing other things. The thing is: I get to use NoteCove myself every day, and can adapt NoteCove to needs I have. I’m certain that the things I need are things you do too as the core items of what these systems need to do is hardly unique.But It’s also been fun and very useful learning how agentic coding works at a real project scale on something that’s not just a toy, because most any process works when things are small, and don’t at all once things get big, and I wanted something big enough to be interesting, and 447k lines of code at the time of this writing, it’s safe to say that I’m there. And lastly, this whole experience helps me to teach others as well. I’ve been doing this for 30 years, and that experience has helped me systematize a bunch of stuff, and now AI is another one. And given the urgency around AI, is why I decided to start Lossy Brain to help teach people about AI workflows and technical (as well as non-technical) leadership.

Where It’s Going

NoteCove is approaching public beta this month. Right now there is only a free tier and it uses your own cloud storage and costs me almost nothing to support. When you need real-time collaboration features or certain forms of integrations, that’s when a) I’d have to build it, and b) for you to pay — because that’s when I’m actually running infrastructure for you. But that day isn’t today, and may never actually be needed. Once I have some users other than me, I’ll find out.

The writer features are coming: progress tracking, timeline visualization, ways to prove your authorship. The task management is there but still being refined. The AI agent collaboration is functional but early. There is an iOS app that has some rough edges, but nothing for Android yet. I’ve got binaries for Mac and Linux, but not Windows (I don’t have access to a windows box right now).

If you’re interested in trying NoteCove, it’s available for download. I’d especially love to hear from writers, knowledge workers frustrated with existing tools, and people who care about data ownership.

NoteCove — A safe harbor for what matters

I spent a half a decade in centralized collaboration infrastructure. Now I want to see what’s possible when we build local-first instead.