Mikipage Blog - Mikipage

Mikipage & Productivity Methods

How Does PARA Map to Mikipage?

Daniel Zhou maps Tiago Forte's PARA method (Projects, Areas, Resources, Archives) onto Mikipage's core features, showing how the product takes over much of PARA's filing work while preserving its mental model:

Projects → Groups: A bounded effort becomes a group workspace. Members write their own notes; group pages synthesize across all contributions with citations back to originals.
Areas → Miki pages: Ongoing responsibilities become smart folders with AI-synthesized overviews — answering not just "which notes belong here?" but "what do I currently know about this?"
Resources → Search, Ask, tags, and Similar notes: Instead of pre-categorizing, capture freely and retrieve by meaning when needed. Structure can wait until a subject earns a dedicated page.
Archives → a flag, not a folder: Archiving is a note property, not a separate container. Archived notes remain searchable and citable; tags can record the reason something was retired.

The net change in daily use: write the note, share it to a group if project-relevant, let pages find it for ongoing areas, search for it as a resource, and archive it when done. Less filing, more retrieval.

Maps of Content That Keep Themselves Up to Date

[2]

Daniel Zhou explains that a Map of Content (MOC) is a structured starting point for navigating notes on a subject, rather than merely a search-result list. While creating a map involves valuable judgment, keeping multiple maps current requires recurring decisions about which notes belong where and how their overviews should change.

Mikipage turns a Miki page into an AI-maintained MOC: users describe the topic and desired structure in ordinary language, while AI finds relevant notes, writes a cited overview, and preserves the original notes unchanged. A note can support multiple pages without copying; users can pin missed notes or refine instructions when a page's emphasis is wrong. The Page view provides the overview and the Notes view exposes its source collection. Background sweeps update pages roughly every 6–12 hours by default, including on the Free plan, while on-demand Refresh uses credits for quicker updates. [2]

Mikipage Philosophy & Vision (continued)

Who Writes the Wiki? From Wikipedia to LLM Wiki to Mikipage

[3]

Daniel Zhou traces the evolution of wikis through three generations to explain what Mikipage is and why it was built. The essay is structured around one question: who writes the wiki?

Traditional wikis (Wikipedia, Confluence): Volunteers write and edit pages directly. This works at Wikipedia's scale but fails for smaller groups — the editing and maintenance burden kills most company, family, or personal wikis within weeks.
LLM Wiki (Andrej Karpathy, April 2026): An AI agent writes and maintains the wiki from immutable raw sources. The maintenance problem is solved, but it requires an AI coding agent, a local machine, and technical setup — and supports only one person's vault on one computer.
Mikipage: Everyone writes their own notes (immutable, never edited by others); AI writes shared pages from all of them, in the cloud, accessible from any browser. Notes map to LLM Wiki's "raw sources," Miki pages to its "wiki," and each page's title and instructions to its "schema file."

The post also covers the group wiki — the part LLM Wiki couldn't do: multiple people each write notes in their own space, nobody edits anyone else's, and AI synthesizes across all contributors into a shared page. The closing line: "Your notes are sacred; the wiki is alive." [3]

Documents, Notes, and Mikipage: Three Different Jobs

[4]

Daniel Zhou explains where Mikipage fits by distinguishing it from both document apps and note apps. Documents (Word, Google Docs) are finished artifacts written for a reader — polished, frozen, presentation-focused. Notes (Apple Notes, Obsidian, Notion) are small, personal, ongoing captures — easy to write, hard to turn into something readable.

The gap neither fills: converting scattered notes into a living, readable page. Note apps store raw material and stop; document apps give you a blank page. Nobody does the part in between — so notes pile up, well-filed and never read. Sharing has the same gap: raw notes nobody reads, or a document you never get around to writing.

Mikipage's answer: notes in, pages out. Notes are yours — written freely, never reorganized by AI. Miki pages are written by AI from your notes, given a title and instructions in plain language, and refreshed as new notes arrive. One note can feed multiple pages without copying. Sharing means everyone writes their own notes, nobody edits anyone else's, and AI synthesizes a shared page from all contributions. The post ends with a comparison table and analogies positioning Mikipage relative to Word, Apple Notes, Obsidian, and Notion. [4]

A Family's Shared Memory Can't Live on One Laptop: Why Mikipage Is Cloud-First

[5]

Daniel Zhou makes the case for Mikipage's cloud-first architecture against the popular local-first trend (exemplified by Obsidian). Three reasons drive the choice: notes available from any browser on any device without installation or syncing; no need to configure plugins and AI models yourself — it all comes built in; and, most decisively, sharing is fundamental to the product, not an add-on.

The core argument is that a group second brain — shared memory across a family, circle of friends, or collaborators — is structurally impossible with local-first tools. Other people can't write to your disk, AI can't synthesize across multiple people's local vaults, everyone's tagging system collides in a shared folder, and fine-grained sharing (private / shared with family / public) becomes manual bookkeeping across multiple copies. Mikipage's model: each note has one owner and one copy; you choose which groups it's shared with; AI synthesizes across all contributors with authorship preserved.

The post also directly addresses the three genuine strengths of local-first — privacy, ownership, and customization — and explains how Mikipage handles each: encrypted storage with protected notes (browser-side encryption for truly sensitive items), full Markdown export and an open API at any time, and an official Obsidian plugin that syncs any Mikipage space to a local folder (giving you the whole Obsidian plugin ecosystem as a workshop, with a live local backup as a bonus). [5]

Mikipage & Productivity Methods (continued)

How Does CODE Work in Mikipage?

[6]

Daniel Zhou maps Tiago Forte's CODE method (Capture, Organize, Distill, Express) onto Mikipage, showing where the human's judgment ends and AI's clerical work begins:

Capture → Notes: Write text-first notes without worrying about folders or hierarchy. Attach files with a short description so AI can search and synthesize them. Voice dictation, MCP integrations, and the Ask chat's "Save chat to note" feature all feed into capture.
Organize → Miki pages, Similar notes, search, and tags: Instead of filing at capture time, organize after the fact. A Miki page acts as both a smart folder and a living readable summary. One note can belong to multiple pages without copying.
Distill → Synthesized pages, Graph view, and Suggest pages: Pages do the repeated compression — turning scattered notes into summaries, timelines, FAQs, or other shapes described in their instructions. Refreshing a page folds in new notes while leaving originals untouched.
Express → Groups and public publishing: Share selected notes into a private or public group. Group Miki pages synthesize across all members' contributions, turning individual knowledge into a shared, citable account.

The net result: capture because meaning is personal, express because audience is personal, and let AI handle the filing in between. [6]

Use ChatGPT to Think. Save What Matters to Mikipage.

[7]

Daniel Zhou argues that ChatGPT conversations are worth saving only when they contain durable insights, decisions, or explanations—not every transient exchange. Moving selected summaries into Mikipage puts AI conversations alongside reading notes, personal observations, and other sources, makes them independent of an AI subscription, and keeps the collection deliberately useful.

The post describes two workflows: ask ChatGPT to summarize a worthwhile conversation and paste the edited result into a Mikipage note, or connect Mikipage to ChatGPT through MCP and use the Mikipage plugin in a Work chat to summarize and save directly. MCP can also search saved notes in later chats, allowing earlier ideas to inform new questions; Miki pages can then synthesize accumulating notes into a continuing topic overview. Saving through MCP uses no Mikipage AI credits, and pages refresh only when explicitly requested. [7]

Mikipage Philosophy & Vision

Why I'm Building Mikipage

[8]

Daniel Zhou traces 25 years of note-taking — from a homemade Microsoft Access database in 2000, through Evernote and Obsidian, to building Mikipage — to explain why he felt compelled to create something new. His core insight: every note app he used was a better filing cabinet, but none ever did anything with the notes. The distilling of raw notes into useful, living knowledge always remained the user's job — and nobody ever gets around to it.

Mikipage's answer is a "miki": a wiki maintained by AI, fed by notes written by one or many people. The ten principles he won't compromise on include: it's a wiki first (not a notebook), a hard line between what you write (sacred, untouched) and what AI writes (alive, always current), everyone writes while AI synthesizes across contributors, powerful retrieval and discovery, write from any browser anywhere, simplicity, frugal AI usage, affordable pricing, full data portability, and non-negotiable security. He's building it for himself and his family, in alpha, in the open. [8]

What I Don't Use Mikipage For

[9]

As a counterpart to the founding vision post, Daniel outlines exactly what Mikipage is not for — arguing that a tool good for everything is just a filing cabinet with more drawers. Mikipage does one loop: write notes → AI organizes them into pages → share those pages.

The six things explicitly excluded are:

Lists & todos — ephemeral by design; use Workflowy or similar.
Long formatted documents — presentation artifacts; use Word or Google Docs.
Photos, voice memos, diagrams — text-first is a deliberate design choice; files can attach but text is the substance.
Research papers & citations — a solved problem; use Zotero.
Development docs — belong in the repo, next to the code.
Calendar events — a coordination protocol, not a note; use Google Calendar.

The dividing line: if it's ephemeral, it's a list; if it's about presentation, it's a document; if a purpose-built tool already owns it, use that tool. What remains — durable, textual, personal writing you'd want your future self to have — is Mikipage's territory. The post also details what does go in Mikipage: life journals, ideas before they're anything, things being figured out, worth-keeping content from others, saved AI conversations, and scanned documents. [9]