So before asking, we thought about the structure.
Rather than focusing on how smart to make AI,
we first looked at when people start speaking, when they stop, and when they bring it up again.
People don't complete ideas at once.
Most start with just one very short sentence.
And that sentence stays in their head longer than expected.
Until they say it out loud.
So the start of this service is always one line.
We don't ask to write long.
We don't ask to explain well.
We just create a state where "it's okay to write down the thought that just came to mind."
From then on, it's a conversation.
But that conversation is not a record.
Nor is it text to be read again later.
It's just a space to let thoughts flow.
AI asks questions in this conversation,
and sometimes offers hypotheses first.
It doesn't always say the right answer.
Rather, it often and deliberately says vague things.
"Usually, there are many cases like this."
"It could be a completely different direction."
This is to allow the person to say "No."
Because thinking becomes clearer in refutation than in agreement.
When the conversation accumulates to some extent, we don't leave it as is.
Instead, AI stops once,
and organizes the thoughts that have come out so far into a structure.
- Who the idea is for.
- What problem it is trying to solve.
- What the solution is.
- Is there a possibility of leading to a business.
- What important parts haven't been said yet.
This structure is not a fixed template.
It changes slightly depending on the nature of the idea.
But there is a common point.
It is filled only with what was actually mentioned in the conversation.
Unmentioned items are left blank.
And we honestly write this:
"Undecided yet."
We decided not to hide this "Undecided yet."
Rather, we decided to clearly show that this is the current state of the idea.
Not as a deficiency, but as a signal indicating it's in progress.
What's important is that this structure does not replace the conversation.
This document is not a conclusion.
It's closer to a snapshot showing how far the conversation has come so far.
So when AI starts the next conversation,
it doesn't try to remember the entire conversation.
It re-enters based on this structure.
"We haven't talked about this part yet."
"Is this assumption you mentioned earlier still the same?"
It acts not as an entity with good memory,
but like a coach who knows the current location.
This process proceeds basically alone.
Because we thought ideas cannot be made public from the beginning.
People are only honest when they are alone.
In a public space, a layer of defense always arises.
So the conversation with AI is strictly private.
It is not shown to anyone.
Even if disclosed later, this conversation itself does not appear.
Instead, what is disclosed is,
the idea structure in its current state, organized through conversation with AI.
And that disclosure doesn't happen all at once either.
At first, you see it alone.
Then you can show it to only a few acquaintances.
If you want, you can share it with the entire community.
Disclosure is not simple exposure, but expanding the scope of conversation.
So the default is always quiet.
Just because you updated, the alarm doesn't ring.
Only when you really want to talk about this version again right now,
the user directly selects and raises it to the stage again.
We placed a small weight on that choice.
Not habitual uploading,
but to make you ask yourself once more if it's a moment really worth talking about.
There is one principle that penetrates this entire structure.
Not to make ideas into 'consumables' that are seen once and passed over.
- Not lining up by views,
- Not evaluating by speed,
- Not comparing based on completeness.
Instead,
making thoughts not disappear,
not stop in the middle,
and be able to continue again.
So we built a service with this structure.
A structure that doesn't evaluate ideas quickly,
doesn't require disclosure from the beginning,
and allows enough time to think alone.
A service that helps you know where your thoughts have come
with AI that doesn't just ask questions
and doesn't just give answers.
We decided to call this service IdeaSynapse.
Because we thought the moment it connects firmly with the world (Synapse)
is more important than the moment the idea flashes in your head (Spark).
After going through this connection, you will be able to say at least this.
"Why this idea is needed in the world,
now I can explain without wavering."
If you have that certainty, anything is possible next.
We created this structure to stand you at that starting line.