I Built the Focus App I Actually Needed. In a Day.

Anchor is a calm, shame-free ADHD focus partner. It notices when your attention drifts to a time-sink site and slides a gentle "back to it?" card in front of you. No task lists, no streaks, no red, no shame. Everything runs on your own machine.

I built it because I needed it. I have ADHD, and what I actually wanted was an accountability partner for myself. Something that would sit alongside me through the workday and notice, gently, the moment I slipped, because I am the person least able to notice it on my own. That is the honest frame for this whole post, so I will say it up front rather than dress it up as market research.

The problem I kept losing to

Here is the shape of a bad afternoon. I sit down with one clear thing to do. An hour later I surface on a browser tab I have no memory of opening, and the clear thing is exactly where I left it. The frustrating part isn't the wandering. It's that I never caught the moment I wandered. The tab that pulled me away felt, at the time, like the task.

People who study this call it time blindness. Drift doesn't get noticed until it has already cost twenty minutes, because "now" rewards are louder than the important-but-later ones. I have read enough about it to explain it well and still lose to it most days. Knowing the mechanism doesn't switch it off.

So most focus tools made it worse. The ones built on streaks and red badges added a second problem on top of the first: now I had drifted and broken a chain, which is just a fresh reason to feel bad and close the app. Punishment is a strange thing to build for a brain that is already reactive to the feeling of failing.

I wanted something that did the one thing those tools skip. Catch the drift as it happens. Say, gently, "back to it?" And then get out of the way.

What I built, and how fast

Anchor is two surfaces sharing one idea.

The first is a Chrome extension. You type your one thing, start a focus session, and it watches the active tab. When you land on a known time-sink, a small card slides in and asks if you meant to be here. No lecture, no counter, no streak to protect. Buoy, the companion voice, is warm about it because a sharp voice is the thing that makes you quit.

The second is a web-app prototype, the same drift-to-nudge-to-re-anchor loop in a page you can open anywhere.

I built the working version in essentially a single day. The action engine ships plain-Node self-tests, nine of them passing, no test framework and no dependencies, because a focus tool that silently breaks its own core logic is worse than no tool. There are two landing directions live on Vercel: Serene, which is calm and breathing, and Quest, which leans into a light RPG flavour for people who focus better with a bit of play. The Chrome Web Store assets are prepared.

That is the part that photographs well. The next two parts are the ones I actually care about.

Privacy as architecture, not a promise

A focus app has to watch what you are doing. That is the job. It also means a focus app is, structurally, a surveillance tool pointed at your worst moments of the day. Most of them handle this by writing a privacy policy that asks you to trust them.

I did not want to ask you to trust me. I wanted the design to make the question moot.

So the extension reads only the hostname of your active tab. Not the URL, not the page, not what you typed. Just "you are on this domain." And it does that entirely on your machine, through the browser's own local storage. Nothing is sent anywhere, because there is nowhere for it to be sent. There is no server in the core loop at all. You could unplug the internet and the thing still works.

That is a different kind of promise. "We will protect your data" is a policy you have to believe. "The data never leaves your laptop" is a fact you can verify by watching the network tab do nothing. When the architecture makes the bad outcome impossible, you don't have to trust my intentions, and I think that is the right way round.

The decision I am proudest of: prove it before charging

There are over a hundred ADHD focus apps. I went looking for the studies showing they help. There is almost nothing. A hundred-plus products, a real condition, and essentially no published evidence that any of them do what they sell.

That is the graveyard I would be adding to. So the roadmap does the unusual thing.

The first backend milestone is not accounts. It is not billing. It is an opt-in, anonymous evidence pipeline whose only job is to answer one question: does using Anchor actually change anything? Not vanity metrics, not daily active users, not a dashboard that makes the founder feel good. Whether re-anchoring after a drift measurably helps the people who opt in. And the plan is to publish the aggregate honestly, including the boring or inconclusive result, if that is what the numbers say.

I want to be plain about why this order matters. Accounts and billing are the milestones that make a product feel real and make it money. Putting evidence first means I find out whether the thing works before I ask anyone to pay for it, which is exactly the moment most products decide they would rather not know. In a category with a hundred apps and zero evidence, "we checked whether it works, and here is what we found" is the only claim left that means anything.

It might turn out that Anchor helps a bit, or only for some people, or not enough to matter. I would rather build the machine that can tell me that than sell confidently into the dark.

The honest status

This is in preproduction, and I would rather you hear the real version from me than infer a shinier one.

It is deployed, but the extension has not yet been loaded and driven in a real Chrome session, so the code is proven by its own tests and not yet by a day of my actual use. The evidence pipeline is a design, not a running system. There is no production database behind it yet. It is not submitted to the Web Store. There are no users. It is a one-day build with a clear plan and a lot of road left.

I am telling you this because the whole point of the app is honesty about whether things work, and it would be strange to launch that on top of an inflated launch post.

Over to you

Anchor is the tool I needed, and it doubles as a portfolio piece, and both of those things are true without cancelling each other out. Building the thing you personally lose sleep over tends to produce better work than building the thing you think will sell.

If focus is your quiet daily fight too, I would genuinely like to know what has and hasn't held up for you, and which tools you have abandoned and why. And if you are building something of your own, especially something where you have decided to measure the awkward truth before monetising it, I want to hear about that most of all.

If you would rather talk it through than type it, my calendar is open.

Frequently Asked Questions

What is Anchor?

A calm, shame-free ADHD focus partner. It notices when your attention drifts to a time-sink site and slides a gentle "back to it?" card in front of you. No task lists, no streaks, no red, no shame. Everything in the core loop runs on your own machine.

Why do streak-based focus apps fail people with ADHD?

They add a second problem on top of the first. You drifted, and now you've also broken a chain, which is a fresh reason to feel bad and close the app. Punishment is a strange thing to build for a brain that's already reactive to the feeling of failing.

What data does a focus extension collect?

In Anchor's case, only the hostname of your active tab. Not the URL, not the page content, not what you typed. It stays in the browser's own local storage on your machine, and there's no server in the core loop, so there's nowhere for it to be sent. You could unplug the internet and it would still work.

Do ADHD focus apps actually work?

There are over a hundred of them and almost no published evidence that any of them do what they sell. That gap is why Anchor's first backend milestone isn't accounts or billing. It's an opt-in, anonymous evidence pipeline measuring whether re-anchoring after a drift helps, with the aggregate published honestly, including the inconclusive result if that's what comes back.

What is time blindness?

The reason drift goes unnoticed until it has already cost twenty minutes: immediate rewards are louder than important-but-later ones, so the tab that pulled you away felt, at the time, like the task. Understanding the mechanism doesn't switch it off, which is the whole argument for a tool that catches the moment rather than lecturing about it afterwards.