Everything so far starts with you saying something. This part doesn’t.
Insights are the background half: Aura watching the threads of your life and preparing what matters before you ask for it. It’s the quietest thing in the product and the one that changes your week most, precisely because it happens while you’re not there.
What arrives unasked
A decision is heading toward you — a job offer, a move, a health thread you mentioned twice — and something arrives already thought about. Not an answer, usually. Context, options, and a recommendation, shaped so that deciding is quick.
The trigger is ordinary conversation. You mention a thing twice; the second mention is a signal. That’s why saying things messily matters more than phrasing them well — the throwaway line about a decision you’re circling is exactly the input that produces something useful three days later.
Deep dives, when the stakes deserve it
For bigger things there’s real research: a Report, built from actual sources, kept in your Library where you can go back to it.
A deep research run costs 16 N — the rate card is on plans — and I’d argue it’s the best value on it. Not because it’s cheap, but because the alternative is an afternoon of your own tabs, and the output is durable: a Report is an object you own rather than a conversation you’ll lose.
Worth doing for the decisions people usually make on vibes. Should I take the offer. Is this the right school. What’s actually involved in the visa. Which of these three is the real choice.
Why it’s paced
You will not get ten insights a day, and that’s a design decision rather than a capacity limit.
Your attention is treated as the scarce resource in the system. There’s a deliberate ladder for reaching you — a quiet card before a push, a push before a call — and a budget for how much of your attention the product may spend. The morning brief carries exactly one suggestion, never a list.
The reason is simple: a proactive product that pages you constantly gets muted, and a muted product is worth nothing no matter how clever it was. Restraint is what makes proactivity survivable.
What you should do with them
Read the one suggestion. It earned its slot against everything else that could have been there.
Say when it’s wrong. An insight that missed is useful feedback, and correcting it in ordinary words is the whole mechanism — memory is yours to fix.
Don’t go looking for more. The temptation is to treat this like a feed and check for activity. Insights aren’t a feed. They arrive when there’s something worth arriving about, and the empty days are the system working correctly.
The honest boundary
Some of this is prediction, and prediction is sometimes wrong. What arrives is prepared thinking, not prophecy — a brief on a decision, not a verdict about your life. You are still the person deciding.
And the same rule from part four governs it: nothing acts on the world off the back of an insight without you at the edge. It can research, prepare, and propose. It cannot send, buy, or commit.
That’s the series
Say it messily · start the threads · give it senses · let Plans run and answer at the edge · and let it work while you don’t.
The whole thing compresses to the sentence I started with: type less, do more. The rest of the product exists so that sentence can be true for someone who has never thought about AI for a minute in their life — which is, in the end, the only version of this worth building.