Self-hosted notes and tasks without running a server

By Gerald · 9 July 2026

Simple personal workspace with a laptop and a physical notebook on a desk, representing notes and tasks running on accounts and infrastructure you control

"Self-hosted" has become a marketing term that covers two very different realities.

In one version, you rent a virtual server, install Docker, configure volumes, set up a reverse proxy, manage SSL certificates, and hope the thing stays up when you go on holiday. You have traded a subscription to a notes company for a part-time job as an unpaid systems administrator.

In the other version, you point a tool at accounts that already belong to you, it provisions the necessary pieces, and your notes and tasks live in databases and storage that are billed to your card and controlled by your login. You own the outcome without having to operate the machinery.

Most writing about self-hosted productivity tools only describes the first version. The second version is quieter, and it is the one that actually matches what a lot of people mean when they say they want to own their data.

What people usually mean by self-hosting

When someone searches for a self-hosted note taking app, they are rarely excited about containers and uptime monitoring. They are trying to solve a different problem.

They do not want their most personal writing and planning to live inside a company that can change its pricing, its feature set, or its continued existence at any time. They want the data to be retrievable even if the original vendor stops answering emails. They want to be able to move the whole thing somewhere else without asking permission.

The technical details of how that happens are secondary. The goal is durability and control, not a particular deployment method.

The problem is that almost every piece of content that ranks for "self hosted note taking app" assumes you are comfortable with the technical details. The recommendations are Joplin with WebDAV, AppFlowy behind Docker, Trilium on a home server, or various single-binary experiments that still require you to keep a machine running and backed up.

These are real options. They are also not what most people are looking for when the phrase "self-hosted" first occurs to them.

The version that does not require operations work

Personal desk with laptop and physical notebook, illustrating notes and tasks you control on accounts you own
Self-hosting your notes and tasks does not have to mean running servers yourself.

There is a middle path that receives almost no coverage.

You create accounts on infrastructure providers you already trust or are willing to trust (a Vercel account, a Convex account, a database provider, object storage). You give the productivity tool permission to use those accounts on your behalf. The tool handles the wiring, the schema, the updates, and the scaling. Your data ends up in resources that are created in your name, billed to you, and accessible to you even if the tool's company disappears.

This is still self-hosting in the sense that matters for ownership. The database that holds your notes is not a table inside the vendor's multi-tenant cluster that they can turn off or raise the price on. It is a database you are paying for directly. The same is true for file storage and any other persistent pieces.

The difference from the Docker route is that you never log into a server, never write a Dockerfile, and never debug a networking issue at 11pm because a container will not talk to its database.

Flow takes this approach. The notes, the kanban board state, the capture inbox items, and any attachments live in a Convex deployment and supporting storage that were created under credentials you control. The software that presents the UI is deployed to a hosting account you also control. If you stop paying Flow tomorrow, those accounts keep running until you cancel them yourself. Your data remains readable and exportable.

The real limitations of this middle path

It is not free. You pay Flow once for the software, and you pay the infrastructure providers for the resources the software uses. For light personal use the infrastructure cost is small, often just a few dollars a month or less, but it is not zero and it is not included in the one-time price.

It is also not the same as running everything on hardware you physically own. Your data is still on someone else's machines, just in accounts that are legally and financially yours rather than the software company's. For many people this is an acceptable compromise. For people who want air-gapped local storage or who distrust every cloud provider on principle, it is not sufficient.

The tool is also more constrained in what it can offer. Because it does not run on a server you fully control at the operating system level, it cannot give you arbitrary plugins, custom database schemas, or direct access to the underlying filesystem for scripting. You get the workflows the tool was built to support, executed on infrastructure you own.

How to evaluate whether this approach is worth it

Ask two questions.

First, do you care more about the data living in accounts under your name, or about the software running on hardware you can touch? If the second matters more, look at the local-first and Docker options and accept the maintenance.

Second, are you willing to pay a larger sum once for software in exchange for never having another productivity subscription? If the answer is yes, then the middle path (one-time commercial tool on your own accounts) is worth comparing against both the free self-hosted route and the familiar subscription tools.

The people who end up happiest with this arrangement are usually those who have already experienced the hidden cost of "free" self-hosted tools (their own time) and the visible, rising cost of subscription tools (their money, every month). They are looking for a durable purchase, not a platform they can endlessly customise.

Frequently asked questions

Does self-hosted always mean I have to run Docker or manage a VPS?

No. That is the most common meaning in current search results, but it is not the only meaning. Some commercial products let you point them at accounts you already own or create, and they manage the application layer while the data resources stay under your control and billing.

How much does the infrastructure actually cost for personal use?

It varies by provider and usage. For a single user with a few thousand notes and occasional attachments, the cost on modern serverless platforms is often a few dollars a month or less. It is not included in the one-time software price, and you should budget for it, but it is usually much smaller than a typical productivity subscription.

What happens if the tool company disappears?

Your data lives in the accounts and projects you created. You can continue to pay the infrastructure bills and either keep using the deployed version or export the data and move it elsewhere. The risk is not that a company turns off a database they control, but that you lose future updates and support for the application code.

Is this the same as local-first tools like Obsidian?

No. Local-first tools store the primary copy on your devices and sync optionally. The middle path described here keeps the primary copy in cloud resources that belong to you. Both approaches can give you ownership. They make different tradeoffs on offline access, multi-device sync, and what happens when a device is lost or stolen.

Can I move later if I change my mind?

Any tool worth considering in this category should make export straightforward and complete. The harder question is whether the destination you move to will feel like an improvement. Many people who leave subscription tools for a one-time owned option do not want to go back to managing their own servers either.

Related reading

My verdict

Self-hosting is a means, not an end. The actual goal for most people is simple: my notes and tasks should not be at the mercy of a company that can raise prices or disappear. The Docker-and-VPS route achieves that goal at the cost of your time. The subscription route gives up on the goal in exchange for convenience. The middle path (one-time software on infrastructure accounts you own) tries to give you most of the ownership benefit with most of the convenience. It is not the right answer for everyone, but it is an answer that almost no comparison article currently discusses in plain terms. If that description matches the tradeoff you have been looking for, it is worth evaluating the small number of tools that actually operate this way.


Read this on flowproductivity.space · More from The Flow Journal · Try the Flow demo