nohmitaina · 脳みたいな
Write the way you always write. nohmitaina distills it into linked concepts, keeps what you return to, and lets the rest fade — the way a brain does. And this brain is shared: your AI agents read it and write back to it, so what each of you learns stays. It grows, and your Core remains.
Like a second brain, quietly at work.
00What is nohmitaina?
One minute on what nohmitaina is — then a look at it in action.
01The Problem
01
Morning journals. Article notes that trailed off mid-sentence. That friction you felt in a meeting. In the moment, every line mattered. A week later, none of it connects to anything inside you. The fact of having written remains. The substance is gone.
02
An ordinary notes app lets you search, re-read, retrace. That isn’t enough. Search returns lines. Not themes. What you actually want is the theme your mind keeps circling back to — and a keyword can’t pull that out.
03
ChatGPT, Claude, all of them. Who you are, what you care about, what you decided last week — you explain it again every time, or it’s simply not there. If your AI already knew you, the first line of the conversation would change.
02The Idea
The brain doesn’t store everything. It is the act of forgetting that lets the important things rise. nohmitaina ports that mechanism into an editor.
When you write, an AI extracts concepts, draws links, scores them, and lets the scores decay over time. What matters keeps surfacing. The rest dims.
This is not an editor for searching. It is an editor for remembering.
03How it works · four moments
A Markdown editor that takes itself seriously. Tables, math, Mermaid, code blocks, link cards — the writing tool you would have built for yourself. Voice in, transcribed by Whisper. No taxonomy to maintain. No filing system to design. You write.
nohmitaina has a thing called Dream. On a quiet schedule, it extracts concepts from what you wrote, drafts a page for each, and weaves them into a structure — broader ideas above, specifics below, related concepts linked across — then re-scores everything. What you wrote about yourself is gathered into one place — the profile your AI reads first. You wake up to a notebook that grew without you.
Turn on the noh CLI, and the agents you already use — Claude Code, Codex, anything that runs a skill — can reach your noh. Ask one to work, and it pulls in your Core and the concepts the task touches, then leaves what it learned behind as a note. Ask one to interview you about your past, and it builds up a picture of who you are, note by note. No copying context in. No starting from scratch. Your brain is simply there, shared, every time.
The concepts you’ve been turning over are already organized. Just open them. The themes, the connections, the surprises — and how each concept’s activation rises and fades over time. You see yourself thinking — and you understand yourself better than you did this morning.
This is what your day could look like.
Download for macOS04The Science Inside
Pillar I
Retrieval strength (how easy it is to recall right now) and storage strength (how deeply it has stuck). A concept can fade from your Core and still live in the deep layers, ready to come back when something connects.
after Robert A. Bjork
Pillar II
Every reference, every revisit, contributes to recall — and decays with time. The Practice Effect and the Spacing Effect, expressed in a single elegant expression.
B(t) = ln(Σ tᵢ⁻ᵈ)
Anderson & Lebiere, 1998 · d = 0.5
Pillar III
Each new concept is checked against everything you already think. If it brings a genuinely new angle, it surfaces. If not, it stays quiet. The longer you write, the more clearly the unfamiliar speaks.
internal: high · medium · low
05What we chose not to build
An all-in-one workspace.
This is not a place to manage tasks, run a CRM, or share docs with a team. It is an editor for one person’s thinking.
A bidirectional linking tool.
You don’t draw the [[brackets]]. The AI finds the concept, makes the page, draws the link. Your job is to write.
A daily-journal app.
No “one page per day” format. Write in the shape that fits your thought. The structure is added later, by the system.
A cloud service.
Your data lives on your disk. As Markdown. We don’t hold it. We don’t need to.
06Pricing
Free
$0 / forever
The full editor. The full loop. Use Claude Code, the Codex CLI, or bring your own API key.
Add-on
$5 / month
Keep your noh in step across the devices you actually think on.
Add-on
$10 / month
Choose the concept pages you want to share with the world. Keep the rest yours.
Launch
If you want the launch page, comments, and a simple way to share it, here is the Product Hunt card.
Editor for thinkers — your past returns when you write
07From the Maker
I journal in the morning. Some days I just talk into the mic and let it run. The writing was never the problem. The problem was that a week later, none of it was still alive in my head.
I tried Obsidian. I tried Reflect. They are good in the way that good tools are good. What I wanted was different: a place where the part of me that is currently thinking would stay visible. A place where I could decide how much of myself the AI was allowed to keep.
nohmitaina is the editor I am building for myself. I use it every day. If any of this resonates, this is for you.
simochan· Tokyo, 2026
Want to talk to me directly? Come over to Discord.Join the Discord →
If this is your problem too.
Download for macOS08FAQ
Where is my data stored?
On your disk, as Markdown. nohmitaina does not push your notes to the cloud. API calls send only the text the model needs to do the work. Sync (when it ships) is opt-in and end-to-end encrypted.
Does the AI rewrite my notes?
No. When an AI assistant writes through nohmitaina, it is saved as an AgentNote — a separate kind of entry, tagged with which agent wrote it. Your own words are never overwritten. You can always tell your thinking from the AI’s.
Can I bring my Obsidian vault?
Drop your existing folder into notes/ and nohmitaina starts ingesting from there. Markdown stays Markdown — your other editors and IDEs keep working.
I don’t have an API key.
Run nohmitaina through Claude Code or the Codex CLI — no billable API key needed. When you want more control, bring your own provider key.
Windows or Linux?
macOS first. Other platforms are technically possible — we ship to them when there is real demand.
Mobile?
Mobile read access is planned as part of Sync. Write on mobile comes after.
Is my data used to train models?
No. Anthropic and OpenAI API policies do not train on API content. Your text is not training data. For the latest details, please refer to each provider’s own policy.
09Releases
| Version | Date | Highlights |
|---|---|---|
| 0.2.0 | 2026-06-25 | The noh agent CLI: AI assistants can read and write your noh. Activation visualizations (sparkline and barometer) and stability fixes. |
| 0.1.7 | 2026-06-12 | Link cards resolve their titles automatically. Hardened security (CSP, SSRF guard on URL clips) and third-party notices. Editor sync and June usability fixes. |
| 0.1.6 | 2026-05-20 | Editor document tabs with macOS titlebar integration. Local Whisper model settings. |
| 0.1.5 | 2026-05-18 | Manual Recall from selected text. Rectangular selection and macro record/replay. Markdown rendering fixes. More resilient Dream processing. |
| 0.1.4 | 2026-05-11 | Multi-workspace activation. Startup catch-up for missed Dream runs. Scoring and Core refinements. |