Hi, I'm Chinat, and I built Habitect. This page is less of a guide and more
of the story: what I am trying to make, why I think it matters, and all the
things I tried before it turned into the app you are using right now.
Pick a topic below to jump straight to it, or read it top to bottom. New to
words like block, graph or agent? Words we use, at the
bottom, explains each one in a line.
Meaningful work
Goals, next actions and activities, and the design they lead to.
I want Habitect to be the last app you will ever have to use. You should
not have to leave it, other than to eat, sleep and hang out with your
friends in real life.
Those are the parts that matter most, so the app should lead you back to
them. What you make here is meant to end up
in real life, not stay on a screen.
We are starting with note-taking. From there it grows into coding and
delegating work to your AI, then texting, and then casual gaming with your
friends. Each of those already has a small home in here.
Why now? AI has made software cheap and quick to make. You can build most
things by describing them in plain words, and most of Habitect was built
that way too.
Having it all in one place matters:
A second brain. You keep better track of your own life. The app becomes a place outside your head that remembers for you.
Sovereignty. Your notes stay yours to control.
Most of what we do in life is note-taking anyway. Note-taking is turning
what is in your brain into something in the outside world.
A caveman did it by drawing and carving on a wall. We do it by writing,
texting and talking. Almost everything we do, talking to each other
included, is a form of note-taking. So a garden for your notes is a good
place for the rest of your life to grow from.
The vision
Habitect is a habitat for you and your agents: a digital garden you both
live in.
You plant a thought as a block, link it to the rest, and come back later to tend it.
Your agents read the same notes, do work on them, and write back right next to where you asked.
Hover or tap a step to read about it.
Habitect in one picture: you write in your notes, hand a block to Chat or an agent, they work from the same notes, and the answer lands back beside the block.Read it as text
Your notes
Plant a thought as a block on a page and link it to the rest. Your agents read the same notes.
Press ⌘J to ask Chat about a block, or ⌘⌃J to hand it to an agent as a task. Keep writing.
Agents work
Your agents work from your notes, so you never explain the whole background again. The Agent log lists what they did each day.
Back in notes
The answer comes back in a card beside the block. Add as notes puts it under the block, grey until you make it yours.
What I care about most is the moment you are in the middle of thinking
something through. You hit a small hiccup, like "what did we decide about
this last month?" or "can someone draft that email?".
Normally you would stop, open another app, explain the whole background
again, and wait. In Habitect you keep going:
Press ⌘J to ask Chat about that block, or ⌘⌃J to hand it to an agent as a task.
The answer comes back in a card beside the block.
Add as notes puts it under the block, inside an Agent result.
Your train of thought never really breaks. And your notes become the
memory your agents work from as well.
You should always be able to tell what you wrote from what your agents
added. So an agent's lines show grey and italic in your notes until you
make them yours. And every day keeps an Agent log of what your agents did.
A lot of AI tools today feel like a slot machine: information keeps
flowing at you, and it is hard to tell which thoughts were yours. I want
the opposite: AI that augments your insight, where you stay the one
deciding what is worth keeping.
From the screen to real life
A screen is where an idea starts, not where it should stay. I do not want
to keep you looking at Habitect. I want what you make here to become things
you can hold, in three dimensions, and time with the people around you.
Hover or tap a step to read about it.
From the screen to real life: an idea in your notes becomes a kit to print, a 3D model, a plushie or an activity with people, and ends up in your hands and at a table with friends.Read it as text
Your idea
A game, an agent, a plan or a thought in your notes.
A trip, a dinner or a game night, planned down to the next small thing to do.
Real life
The screen goes away. You play at a table with friends, go on the trip, or hand over what you made.
You can see it in the app today:
Things you can hold. Add #bring-it-to-real-life to a game, an agent or an idea and you get a kit to print and play at a table, a 3D model of its little character, and a pattern to sew it as a plushie. It opens at Sprout, after three days of writing in a row. See Bring it to real life.
Friends in one room. In the Habiverse, everyone gathers at one screen, or a projector, and each friend's phone becomes their pad.
Plans that get you moving. Every goal comes down to a next action, the one small, physical thing you can do right now. See Activities.
This matters most for games. At a table with friends there is no feed
pulling you back, so the part of a screen that hooks you falls away, and
what is left is the fun of playing together.
Some of this is not here yet. I would like kits to be printed and posted to
you, figures and plushies to be made and sent, and cards a phone can read,
so friends play in person and the screen is only the referee.
What I believe
Everything we do, and every decision we make, is a reflection of who we are.
Our values and beliefs drive our decisions, and our decisions drive how we
design a product, from what a button says to who gets let in.
Habitect is the same. One of my core values is effective generosity:
being generous in a way that lasts, for you and for the people around you.
It guides the philosophy behind the app and the choices we make in it. Its
axioms are:
Always listen to your inner child. Meet your own real needs first, so that ego, insecurity or anxiety are not what drive your choices.
Give authentically and optimistically. Give with no strings attached, and assume the best of people.
Generosity must be progressive. Give more as trust is shown, rather than everything to everyone at once.
You can’t give what you don’t have. Time and energy run out too, so keep enough for yourself to keep giving.
Under them sit a few ideas I lean on all the time: play an infinite game
(grow the pie rather than fight over it), think long term, start from first
principles, and be radically transparent, even when the honest answer is
"I do not have time".
You can see them in the app:
The little plant that greets people without an invite lets them in as they show trust, not all at once.
Agents start small and earn more as they prove themselves.
A lot of Habitect helps you look after yourself first. A gardener running on empty cannot tend anything.
The whole philosophy, with the books and thinkers behind it, lives in the
effective generosity wiki, which keeps
growing as I learn.
Meaningful work
I think most of the stress in work comes from big, vague things sitting in
your head with no first step. So Habitect gives meaningful work a simple
shape, a bit like a plant:
A goal you actually care about, the reason the work matters.
Sub-goals that break it into parts, like branches.
The next action: the one small, physical thing you can do right now.
Activities: how that action gets done, as a focus session, a checklist or a block of time on your calendar.
That shape drives a lot of design choices:
Tell Chat a goal and it breaks it into sub-goals and timed activities for you (see Activities).
Your first few minutes in the app end with "what is one thing you want to get done today?" and then "what is the next action?".
Focus sessions are front and centre, rather than another feed to scroll.
Agents take on the busywork, so the part left for you is the part that needs you.
Designed for your mind
Habitect is built on what we know about how people think and form habits,
from research on memory and planning to research on motivation. A few
examples, each with its research in Sources:
Get it out of your head. Your working memory only holds a few things at once (Cowan, 2001). Writing a worry down as a block frees it up, which is why capturing a thought is always one keystroke away.
Close the loop. An unfinished task keeps nagging at you, until you make a plan for it (Masicampo and Baumeister, 2011). Naming its next action, from your first few minutes on, turns the nag into a plan.
Start tiny. Tiny habits stick where big resolutions don't (Fogg, 2019). Your stages, Seed to Sapling, grow from writing a little on a few days in a row, not from long sessions.
Show progress. Small wins keep you going (Amabile and Kramer, 2011), so finished steps, stages and the Skill tree show how far you have come.
One thing at a time. Too much to learn at once crowds out the thinking (Sweller, 1988). New tools open as you grow, rather than all on day one, so the app never asks you to learn everything at once.
Your own words. People care most about what feels like their own choice (Ryan and Deci, 2000). From your first few minutes, the app asks about you and uses your answers, so it feels like yours, not a script.
Well-being is part of the design too, not an extra. On your Home screen:
Tool
For
Calm
A minute of breathing
Habits
The small things you want to keep up
Focus
One thing done well
What it avoids matters as much: no endless feed, no slot machine of
notifications. AI waits to be asked or shows its work in grey, so you stay
the one deciding.
Sustainable, honest and open
I want Habitect to create value that lasts, for you, for the people who
build it and for the planet, rather than value squeezed out of your
attention or your data.
Your data is yours. We do not sell your data and we do not use advertising trackers. The AI providers we use don't train their models on it. Your notes stay readable and exportable, so you can always take them with you.
Zero data retention is where we are going. The goal: nothing you write is kept anywhere you didn't choose. It is planned, as part of a Believer plan for early believers, but it is not live yet. I would rather tell you that than pretend.
What we keep today. What Habitect needs to run: your notes and files. During the closed beta, also recordings of how the app is used, which you can turn off in Settings → Privacy. The Privacy Policy lists everything, in plain words.
You can always see what AI did. Agent lines are grey and italic until you make them yours, and the Agent log on each day lists everything your agents did.
The cost of AI is shown, not hidden.Your AI footprint shows how much energy your AI used this month and how to offset it.
Out in the open. How the company is owned and funded is written out in full (see Investing), and the app is built in the open from the reports you file.
Where I see the world going
We are moving from one "super AI" towards everyone having lots of agents,
tools and apps working for them. Right now that is pretty chaotic.
Your context is what you know, what you are working on and who you work
with. It is scattered across email, calendars, notes, Notion, messages,
voice memos and browser tabs.
Every time you start a conversation with a new agent, you copy-paste your
life back into it. Each platform keeps its own memory, and none of them
really have an incentive to share it with you.
So the question I keep asking is: what if your memory was yours?
You could give your context to any agent you choose, as easily as plugging in a memory stick.
It would live somewhere you can actually read, edit and grow.
The people and teams who do well in the next few years will be the ones
whose agents really know what they know. I want that to be something you
own, in plain notes, not locked inside someone else's product.
Trust gets built gradually. Agents will make mistakes, like sending an
email when you only asked for a draft. The answer is not to hand them
everything on day one. Start small, correct them, and let them earn more
over time.
It is a bit like how cars went from helping you stay in your lane towards
driving themselves. That kind of slow, visible growth is what gardens are
good for.
What I envision
A garden that is yours. Your notes, your agents' skills and their memory live together, in a form you can read, export and take with you.
Agents as gardeners, not a feed. You write, they work, and both of you read the same notes. You decide what stays.
Every source flowing in. Your calendar, email, meetings and the notes you already have arrive in your journal on their own, so you never rebuild your context by hand.
Skills you write once. The way you like things done is written down once as a skill, a short playbook, and every agent you use follows it, wherever it runs.
Gardens that connect. Share a page, a project or a whole graph with the people you build with, and one day let gardens talk to each other.
A garden that reaches real life. What you grow here turns into things you can hold and time with friends, away from the screen.
The experiments
Habitect did not come out of nowhere. It is basically the latest attempt at
the same question I have been circling for over a year, which is how do you
give people and their agents one shared context. Here is what we tried.
Aug 2025: Contextus. Every project got one shared memory. Every tool we built read from and wrote to it: an idea generator that wrote a product plan, a website builder, a slide maker, a form filler for hackathon entries.
The first time it ran end to end in a live demo, one idea became a plan,
a website and a slide deck, all tied to the same project. That was when I realised the shared memory was the product, not
any one tool.
Sep 2025: the first Habitect page. I wrote it up in Notion as "Habitect: AI Agent Digital Garden", a second brain where your agents, tools and knowledge connect: your own notes and your agents' notes in one place. The vision was already there: every person or team with their own living garden where agents grow, adapt and work together.
Mar 2026: Habitect as a memory. I rebuilt it as a folder of notes and small programs, in three parts:
One brought in my notes, email, calendar, files and messages from the apps they lived in.
One held what I know, sorted by area of life.
One let agents act on it.
I also kept my notes apps and calendar in step with each other, and
brought in my LinkedIn contacts.
Apr 2026: my notes with a history. I started keeping every past version of my notes, saved each day, and kept my agents' instructions and memory there too. That way I could see what changed, go back, and take it all from one computer to another.
May to Aug 2026: snapshots and self-tracking. Regular snapshots of what my agents know, an Agent log that copies every line my agents write into my daily note, and screen-time tracking to see where my attention was actually going.
Sep 2026: the Gate. At the Own Your Intelligence hackathon, a weekend building contest at Y Combinator, I tried teaching a tiny AI from my own browsing to be a bouncer for my attention, answering one question: should I be spending time on this right now?
We didn't place. But it taught me that the labels you give your own data
are what make an agent really yours.
Sep 29, 2026: Block Outliner. I started building my own outliner from scratch: a notes app where every line is a bullet you can fold and move.
Two days later I wrote my daily journal inside it and was honestly amazed
that it worked. Then I brought in my old notes, over nine thousand pages,
which promptly knocked the database over.
How we got here
Block Outliner started as an outliner and nothing more: pages, blocks,
journals and links both ways. It looked and flowed like the notes app where
I have done my best thinking for years.
But as soon as I was writing in it every day, the old Habitect ideas came
straight back in:
A button to report a problem.
A way to hand a block to an agent without leaving the page.
Agents picking those tasks up, writing their answers back under the block, and logging what they did in the day's Agent log.
On the night of Oct 1, 2026, it stopped being Block Outliner and became
Habitect, and moved to its own home. The next day we added the things the
original vision always had in it:
skills that every agent shares
your calendar and email flowing into your journal
new graphs that bring in your notes from another app
a spare computer that keeps agent tasks running while yours is closed
public notes you can share as a small site
plugins
The part I find most fun is how it gets built. Almost every feature began
as a report someone filed from inside the app. A team of coding agents then:
picked each one up,
wrote the checks it had to pass,
built it,
and sent it for review before it went live.
The app saw more than 250 changes in its first five days. This page started
the same way, as a report asking for exactly this story to be written down.
So Habitect is, in a small way, already the thing it describes: a place
where I write what I want, and my agents and I grow it together.
I wrote up the idea behind this way of building, and how you can join in,
in Agentic engineering.
If you want to see the pieces at work, start with
Agent tasks, Data sources and
Plugins. And if something feels off or you have an idea,
file a report. It will probably end up in here.
Investing
I want the people who use Habitect to be the people who own it.
Invest in Habitect writes out how that will work:
who can invest, and how much
where the money goes
every share of the company
a calculator, the policies it keeps to and how the plan came together
It is a plan, not an offer: no money is taken yet.
Words we use
Habitect is a garden, so a lot of its words come from gardening. Here is
what each one means, in one line.
Block: one line of writing, with a bullet in front. Blocks can sit under other blocks, like branches on a stem.
Page: a list of blocks with a title. Linking a page that does not exist yet makes it.
Daily note: the page for today. A new one opens each morning, so you always have a fresh bed to plant in.
Graph: all your pages together, joined by their links. You can have more than one, say one for work and one for home.
Tag: a link written with #, like #idea. It groups every block that uses it on the tag's page.
Action: a special tag that sets an agent to work, like #create or #bug. Write it on a block, leave the block, and it acts. Actions used to be called commands.
Duplicate: copies an agent under a new name and colour.
Clone: builds an agent for a role from a conversation, without copying the person.
Agent: an AI helper that works in your notes with you. Its writing shows grey and in italics until you make it yours.
Activity: something you do in life, like journaling, exercising or deep work, with goals, sub-goals and tasks. A page's template and timer support it.
App: something you use or make with its own screen, like Workouts, a game or a story. Apps, activities, actions and agents are the 4 As.
Seed: the first stage of your account, where everyone starts.
Chat: a conversation with an agent, in a panel beside your notes. Ask a question and get an answer right away. ⌘J asks Chat about what you selected.
/Ask: a question typed into a block. Unlike Chat, its answer comes only from your own notes, and it is written on the page under the block, showing where each part came from. Chat is a conversation beside the page; /Ask leaves its answer in your notes. The ⌘P item that opens Chat is Chat with your agent's name.
Task: a piece of work you hand to an agent. Its answer comes back in a card beside the block you sent it from, and
Agent result: the heading an agent's answer sits under once you press Add as notes. Its lines stay grey until you make them yours.
Agent log: a folded list on your daily note, with a short line for each thing your agents did that day.
Chief: the agent that leads the others. It splits a big job into steps, hands each to the agent that fits, and puts the answers together.
Skill: a written playbook an agent follows for one kind of job, like a recipe card. Skills belong to agents.
Ability: something you have learned to do in Habitect, like nesting blocks. The Skill tree is the map of your abilities.
Workflow: a saved task you start with one click.
Loop: a workflow with a schedule, like watering every morning.
Connector: a link from Habitect to another app or service your agents can use, set up under Agents & connectors.
Runner: a program on a computer that waits for an agent's tasks and runs them, on Habitect's server or on one of your own computers. One computer can have several.
Stage: how far you have grown: Seed, Sprout, Seedling, then Sapling. Writing a few days in a row moves you up, and each stage opens new tools.
Agent growth: an agent's two growth states, Seed → Sprig. Every agent starts as a Seed, and after many uses it evolves into a Sprig, shown by a white core in its dot. An agent's growth states are separate from your account's stages: an agent grows by being used, and you grow by writing.
Sprig, Shrub, Tree: how a page grows in All pages, Sprig → Shrub → Tree, as you write on it.
Garden: where your agents live. Visit them, see what they got done, and play the garden games. It is in the left sidebar.
Plugins: extra tools for your notes, like a habit grid, and the place to connect your other accounts. It is in the left sidebar too.