
The moment that broke us
Adora and I were sitting in front of a laptop at 11pm, replying to a difficult client message we had been putting off all day.
We typed the whole thing out. Hit send.
Immediately regretted it.
Not because we said anything wrong. But because we said it wrong, from a place of frustration, not strategy. We were emotionally activated, and we made a decision we already knew we should not make.
The worst part? We had made this exact mistake before. More than once. We knew better. We just could not slow ourselves down fast enough in the moment.
That is when Adora and I started asking a different question.
Not "how do we get better at this?" But: what would make us keep failing at this?
What is inversion thinking?
Most of us are trained to chase the goal forward.
How do I succeed? What should I do more of? What is the next step?
Inversion thinking flips that entirely. Instead of asking how to win, you ask: what would make me fail? Then you remove it.
Charlie Munger, Warren Buffett's partner, built a $300 billion business partly on this idea. He said:
"Invert, always invert. Many problems cannot be solved forwards. They can only be solved backwards."
It is deceptively simple. Map the failure first. Clear the path. Then move.
The problem is that almost nobody can do this in real time.
Not when a difficult client is waiting for a reply. Not when someone offers you a vague collaboration and your excitement is already outrunning your judgment. Not when you are scattered across twelve tasks and your brain is screaming that everything is urgent.
In theory, we know to think inversively. In practice, under pressure, the default brain takes over, and the default brain goes forward, not backwards.
That gap, between knowing the framework and using it when it actually matters, is what Twinkle was built to close.
Meet Twinkle
Twinkle is an AI Focus Agent.
You bring it a situation. A decision. A person. A feeling. Even just one word: stuck, client, money, angry. Twinkle runs the inversion loop in the background. It maps what is really going on, filters what is noise, and gives you one short, honest reply.
Not a wall of advice. Not a 12-step framework. Just the next right move.
It works because most of the thinking you need in a hard moment is not complicated. It is just unavailable, buried under emotion, urgency, and the pressure to do something right now.
Twinkle creates one second of structured pause. That second is everything.
What it looks like
You type: possible collaborator
Inversive Thinking Framework
Got it. Treat this like a vague opportunity. Do not commit until they make it concrete. First: get the goal, roles, timeline, and value exchange in writing.
You type: angry reply
Emotional Slowdown Framework
Got it. Treat this like an emotional reply. Do not send it yet. First: write the unsent version. Get it out of your system. Then decide what to actually send.
You type: should i say yes
Noise to Signal Framework
Got it. Treat this like a decision under pressure. Do not decide until the main risk is clear. First: write the outcome you want, then write what would make this go badly.
One input. One framework. One first action.
That is it. That is the whole product.
Because if Twinkle gives you twelve things to think about, it is just adding more noise to the noise.
What we actually built
The Inversion Loop Engine: a background reasoning process that clarifies your real goal, maps the most likely failure path, and filters out everything that does not move you forward. It runs silently on every input, even a single word.
One-Bubble UI: built in Next.js. No account. No dashboard. No settings to configure. You type. Twinkle replies. The input clears. You move.
Vague Input Intelligence: Twinkle does not need a full sentence. It handles signals like stuck, work, boundary, money, and infers the likely need. Because when you are in the middle of a hard moment, you are not going to write a paragraph. You are going to type one word and need something useful back.
High-Stakes Safety Layer: legal issues, medical situations, mental health topics. Twinkle does not treat these like productivity problems. It slows down, acknowledges the weight, and points you toward the right kind of help.
7 Named Frameworks: every reply tells you which framework it is using. Not just what to do, but how it is thinking. So over time, Twinkle does not just give you answers. It teaches you a new way to think.
The seven frameworks: Noise to Signal · Inversive Thinking · First Principles · Human Risk Filter · Boundary Framework · Emotional Slowdown · Daily Execution.
Why this matters
Adora and I are not developers. We never were.
We built Twinkle because we were tired of knowing the right frameworks and still not using them when it counted. Tired of making avoidable mistakes, not because we lacked information, but because we lacked structure in the moment.
Most people do not fail because they do not know enough. They fail because they say yes too fast. They trust the wrong people. They react when they should pause. They chase noise disguised as opportunity. They make expensive decisions from an emotional state they did not even notice they were in.
Inversion thinking is the antidote to all of that. And Twinkle is how you make it a daily habit, not a concept you read once in a book and immediately forget.
The goal is not to think less. It is to think smarter. One decision at a time.
Try it
Built by Wan Wei and Adora. Copy written with Claude AI.
Women in Claude