Context is the whole game now. Here is how to stop losing yours.

The best developers and writers I know are not faster typists. They are just better at not losing their place.

That is the quiet skill behind good work: holding the right handful of things close while you use them. The reference image next to the design. The error message you need to paste into your AI assistant. The link you will want again in ten minutes. Lose that thread and the work does not just slow down, it fragments. You spend the afternoon rebuilding a context you already had this morning.

I built Tansei because I kept watching that happen to myself, and to everyone I showed it to. This post is not really about the app. It is about why keeping your context close has quietly become the most important productivity skill of 2026, and a few things you can do about it today, whatever tools you use.

What is context, and why do you keep losing it?

Context is the small set of things you are actively working with right now. Not your whole file system, not your entire brain, just the working set: the five tabs, the three screenshots, the snippet, the color, the note to self.

You lose it because the working set lives nowhere. It is scattered across your clipboard (which holds one thing), your browser tabs (which multiply), your downloads folder, a Slack message you starred, a screenshot on your desktop named Screenshot 2026-08-24 at 14.02.11. Nothing gathers it. So every time you switch apps, you pay a small tax to find your place again.

How much does that actually cost?

More than it feels like, because the cost is paid in seconds, hundreds of times a day.

• The average digital worker toggles between apps and websites about 1,200 times a day, which adds up to roughly four hours a week just reorienting. That is close to five working weeks a year. (Harvard Business Review, 2022.)
• After an interruption, it takes on average 23 minutes and 15 seconds to return to the original task. (Gloria Mark, University of California, Irvine, in Gallup.)
• Knowledge workers spend up to a fifth of the workweek, nearly a full day, simply searching for information they need. (McKinsey.)

None of these show up on an invoice. They show up as a vague sense that the day disappeared and the real work barely moved.

Why did AI make this worse instead of better?

This is the part people do not expect. The AI tools that were supposed to save time have made context loss sharper, not softer.

If you code with an assistant like Claude Code or Cursor, or you write with one, you already know the loop. You gather the ingredients the model needs: the screenshot of the bug, the exact error text, the relevant snippet, the design you are matching. You paste them in. The model helps. Then the session moves on, the context window fills, and the next time you need those same ingredients you are hunting for them all over again.

Fifty-one percent of professional developers now use AI coding tools every single workday (Stack Overflow Developer Survey, 2025). That is a lot of people spending a lot of their day feeding context into a text box. The typing is not the slow part. The gathering is.

Context engineering: the idea worth stealing from the AI people

Here is the concept that reframed all of this for me.

In 2026, the skill everyone building with AI talks about is no longer prompt engineering. It is context engineering: deliberately curating the information you put in front of a model so it can actually do the job. Anthropic frames it as maintaining the right set of tokens for the model to attend to. The reason it matters is that models have a finite attention budget. As the context grows noisy, precision drops and the model starts missing things it should have caught.

Now read that last sentence again and notice that it is also true of you.

You have a finite attention budget too. When your working set is scattered across twelve places, your own precision drops. You miss the thing you already had. The whole discipline of context engineering, curate the working set, keep it close, make the implicit explicit, is not really an AI technique. It is a human one that we are finally naming because machines made the cost impossible to ignore.

So the useful move is to do for yourself what you already do for the model: give your context a home.

How to keep your context close (whatever tools you use)

A few habits that help, none of which require my app:

1. Separate capturing from organizing. The moment you stop to file something is the moment you lose your flow. Capture instantly, sort later or never. Most things you keep for ten minutes, not ten months.
2. Give the working set one edge of your screen. Not a folder you open, not an app you switch to. Somewhere always visible, so your context sits beside your work instead of underneath it.
3. Make implicit context explicit. The note that says “use this one” next to a color, the tag that groups three references together. Your future self, and any AI you hand it to, cannot read your mind. Write the reason down.
4. Stage your AI ingredients before you prompt. Before a session, gather the screenshot, the error, the snippet in one place. You will prompt faster and re-explain less.
5. Archive instead of deleting. Deleting is a decision, and decisions are friction. Move things out of sight with an easy undo, and you will keep the shelf clean without ever hesitating.

Where Tansei fits

Tansei is a shelf for your desktop. It sits docked at the edge of your screen and quietly keeps the things you copy: text, links, images, files, colors, and code, saved automatically the moment you copy them. You can also grab a screenshot, pick any color on screen, jot a note, or record a voice memo, and it all lands in the same place.

The part people fall for is dragging back out. Pull anything off the shelf straight into your browser, your editor, a document, or a prompt. That is the one thing Windows clipboard history cannot do, and it is the difference between a list you scroll and a working set you actually use.

Everything stays on your machine. There is no account and no subscription, just a one-time purchase. I made it that way on purpose. Your working context is the most personal thing on your computer, and it should never have to leave it.

Tansei began on Mac, where it lives on the Mac App Store. Today it comes to Windows too, certified and available on the Microsoft Store. So whichever machine you work on, the shelf is there. It is free to install on both, and one purchase unlocks the full app for good.

Get Tansei for Windows and Mac

We have spent this whole decade teaching our machines to hold context: to remember the thread, to pick up exactly where they left off. It is worth asking why we never expected the same for ourselves. The goal was never to do more in a day. It was to stop losing the day to the small work of finding your way back.

Frequently asked questions

What is context engineering?

Context engineering is the practice of deliberately choosing and organizing the information you put in front of an AI model, or yourself, so the work gets done well. For AI it means curating the right tokens within a finite attention budget. For people it means keeping your active working set close and clearly labeled instead of scattered across tabs, folders, and clipboards.

How much time do people lose to context switching?

Research suggests a lot. Digital workers toggle between apps around 1,200 times a day (Harvard Business Review, 2022), it takes about 23 minutes to refocus after an interruption (Gloria Mark, UC Irvine), and up to a fifth of the workweek goes to searching for information (McKinsey).

Is Tansei just a clipboard manager?

It includes a clipboard history, but it does more. It keeps images, files, colors, screenshots, notes, and voice memos on one shelf, and it lets you drag any of them back out into another app, which standard clipboard history cannot do.

Does Tansei send my data anywhere?

No. Everything you save stays on your machine. There is no account and no subscription, just a one-time purchase.

Is Tansei available on both Mac and Windows?

Yes. Tansei began on Mac, where it is on the Mac App Store, and it is now available for Windows on the Microsoft Store. Each platform is a native app built to feel at home on that system, and each is a separate one-time purchase.

What do I need to run it?

On Windows, Windows 11 (64-bit), from the Microsoft Store. On Mac, from the Mac App Store. It is free to install on both, with a one-time purchase to unlock the full app.