What I Don't Use Mikipage For

Author: Daniel Zhou · Date: August 6, 2026

In my [1] I wrote about why I'm building Mikipage after 25 years of using other people's note apps. This one is the other half of that argument, and honestly the more useful half: what I don't put in it.

I'm the founder. I'm supposed to tell you my product is good for everything. But an app that's good for everything is a filing cabinet with more drawers, and I already own several of those. The only way Mikipage does one thing well is if it refuses to do the other things — not "hasn't gotten to them yet," refuses. So here's the list, and the reasoning behind it, so you can decide in about two minutes whether this is a tool for you.

The one thing

Mikipage is a purpose-built, opinionated web app that does exactly one loop:

You write notes. AI organizes them into pages. You share those pages with the people you want.

That's it. Three moves, and each one is doing a specific job:

Write is deliberately dumb. No folders, no vaults, no "where does this go." You type, you save. Organizing at capture time is the tax that kills note-taking, and I'm not charging it.
Organize is AI's job, not yours. A Miki page is a living wiki article synthesized from your notes — rewritten as more notes land underneath it. You never file anything. You never re-read four years of journals to find the pattern; the page does that and keeps doing it.
Share is why it's a wiki and not a diary. Your notes stay yours — nobody else edits them, ever. But they can feed a group's pages, so my wife's notes and mine become one coherent page about the thing we're both thinking about, with no copies and nothing to keep in sync.

Everything about the product falls out of that loop. Which is also how I know what to say no to: if something doesn't help you write, doesn't help AI make a better page, and doesn't help you share a page with someone — it's not going in.

Six things it isn't for

Ordered roughly by how often people ask.

1. Lists. Todos, groceries, packing lists.

This is the most common request and the clearest no. Lists are ephemeral by design. You write them so you can cross them out, and once everything's crossed out the list has done its job and should evaporate. Nothing about it belongs in a long-term memory — nobody needs an AI page synthesizing four years of grocery runs into "themes in your dairy purchasing."

Lists also want a completely different interface: drag to reorder, indent to nest, check to kill. That's a different app, and a good one already exists. I use Workflowy and have for over ten years. I'm not going to build a worse Workflowy inside a wiki.

2. Long formatted documents. Term papers, theses, business proposals.

These are presentation artifacts. Their whole point is how they look to a reader — page layout, headers, citations in a specific style, track changes, comments in the margin, a PDF at the end that has to look exactly right. Mikipage is Markdown text, because Markdown text is the substance AI can actually reason over. The moment I add page layout and style controls, I've built a worse Google Docs.

Write those in Word or Google Docs. What does belong in Mikipage is the thinking that feeds them: the raw notes, the half-formed argument, the reading you did. Get the document out of that; don't try to compose the document here.

3. Photos, voice memos, diagrams, handwriting.

Mikipage is text-first, and that's a design decision rather than a gap in the roadmap. Text is what AI organizes well and what stays readable in twenty years. Each of the other modalities is a real craft with real apps behind it — your camera roll, a voice memo app, Excalidraw or Figma, GoodNotes on an iPad. I'm not competing with any of them.

The nuance: files can ride along as attachments on a note. That's how I keep scans and PDFs (more below). The rule is that the searchable, synthesizable substance is the text you write around the file — not the file itself.

4. Research papers and citations.

Reference management is a solved problem with decades of purpose-built software behind it: DOIs, BibTeX export, PDF annotation, a Word plugin that formats your bibliography in whatever style the journal demands. I use Zotero, and before that Mendeley, and before that EndNote. All three are better at this than a general note tool will ever be.

Again there's a split worth naming: the library lives in Zotero; what I think about a paper lives in Mikipage. Those are genuinely different objects, and conflating them is how you end up with a reference manager full of prose and a note app full of metadata.

5. Development docs.

Design docs, API references, runbooks, architecture notes. These want to live next to the code, in the repo, in the monorepo, reviewed in the same pull request that changes the behavior they describe. Documentation that drifts out of sync with the code is worse than no documentation, and the only reliable cure is keeping them in the same commit. GitHub already won this one.

6. Booking an event on a calendar.

A calendar event isn't a note, it's a coordination protocol: a specific date and time, invitees who need to accept or decline, reminders that fire, time zones that have to resolve correctly, and a free/busy check against everyone else's schedule. None of that is writing, none of it is something AI needs to synthesize into a page, and none of it is shareable in the way a Miki page is shareable. It's Google Calendar's job, and it already does it well.

What does belong in Mikipage is the thinking around the event — notes on what you want to cover in the meeting, or what happened after. The event itself lives on your calendar; put a link to it in your notes if you want the two connected.

What I actually put in it

Here's the flip side, and it's a much longer list. The test I use is simple: would I want a version of me six months from now to have read this? If yes, it goes in Mikipage. Not exhaustive, but this is genuinely what's in mine:

Life as it happens — daily journals, reflections, lessons learned, parenting tips, family planning, travel planning, career thinking.
Ideas before they're anything — entrepreneurial ideas that showed up in the shower, research ideas, half-thoughts I'd otherwise lose by Thursday.
Things I'm figuring out — financial management learnings, health notes, technical tidbits, what I'm learning about using AI effectively, board game strategy.
Things worth keeping from other people — quotes, book reviews, recipes I actually tried and what I'd change, interesting articles I want to hold onto. (I usually have Claude extract the substance of an article and write it in via MCP rather than clipping the page — that's its own post.)
Conversations with AI worth saving. An hour of good back-and-forth that produced something real, dropped in as a note instead of scrolling away in a chat history.
Documents — scans of receipts, tax paperwork, signatures, personal records; PDFs as attachments with a few lines describing what's in them so they're findable later.

Look at what those have in common. They're all text I wrote in my own words, they're all things I'll want back at some unpredictable future moment, and I have no idea today what category any of them belongs to. That last part is the whole point: I don't have to know. Six months of scattered recipe notes become a page about how I cook. Four years of journals become a page about what changed. Nobody filed anything.

And roughly a third of it is stuff I eventually want someone else to see — my wife, my kids, a couple of friends — which is the part no other tool in this category does at all.

The dividing line

Strip it down and there are three questions:

Is it ephemeral? Then it's a list, and lists belong somewhere you can cross things off. Mikipage is for things you'd be annoyed to lose.

Is it about presentation? Then it's a document, and documents belong where the formatting lives. Mikipage is for substance.

Does a purpose-built tool already own it? Then use that tool. Citations, code docs, images, audio — each has decades of specialized work behind it. Mikipage's job is your own writing, and it's the one thing nothing else is quite doing.

What's left after those three cuts turns out to be most of what I actually want to keep: the durable, textual, personal stuff. The second-brain material. It's a big category, and it's been the worst-served one for 25 years — because every tool in it is a filing cabinet, and I never once got around to doing the filing.

The other reason to say no this clearly is that the product stays small. Every feature I don't build is a screen you don't have to learn and a decision you don't have to make before you can type a sentence.

Storing it, though, is only half of it. The reason any of this matters is what you can do with it afterward — search, asking questions across everything you've written, and the pages that write themselves out of it. That's the next post.