The evening vote. One tap a day, sealed instantly.
Each evening of a running decision, Counsel.day asks you one question and gives you one tap. You may add a sentence. You may add a paragraph. You may write nothing at all. Whatever you record is sealed the moment you commit it, and stays sealed until the final day of the period you chose at the start. This is the smallest possible interaction the product has, and it is the only one that runs every day.
A single question, a single tap, an optional sentence.
When the evening prompt arrives, the surface is deliberately quiet. The question you posed on day one is shown back to you in full. Four options sit underneath it: Strong Yes, Lean Yes, Lean No, Strong No. Beneath the buttons is a single optional note field. There is no graph, no rolling tally, no peek at your partner's drift. The whole interaction is built to fit inside the last minute before you go to bed, and to leave no surface area for second-guessing.
The four-band scale is the only honest answer to a hard question. A pure Yes/No flattens the day you are actually having; a slider invites you to perform precision you do not have. Strong Yes versus Lean Yes is the question every couple actually argues about, on the evenings the disagreement is most real. The four bands surface that, and the analysis on the final day reads them as a trajectory rather than a verdict.
Cast your vote for tonight. The mood you are in right now is exactly the signal the analysis needs. If you feel different tomorrow, the vote tomorrow will say so.
The moment you tap, the vote is sealed.
The sealing is not a UI flourish. It is an architectural fact. The instant you commit, the vote and any note you wrote are encrypted with a key bound to the verdict date and written to a sealed-read row in the database. There is no admin override. There is no support escape hatch. Not even the founder of the product can look at a sealed vote before the reveal day. The privacy is enforced at the level the data lives, not at the level of what the interface chooses to show.
The reason this matters is the reason the whole product exists: if either partner could see how the other was drifting mid-period, the votes that followed would not be honest. Sealing is what makes the season of voting work. The privacy is the mechanism, not a feature.
The deliberate constraint of a single evening prompt.
The product gives you one prompt a night for the duration you chose, and no more. You cannot vote twice on a Tuesday because Tuesday only happened once. You cannot revise yesterday's vote because yesterday was a real signal and revising it would erase exactly what the analysis needs to read. The pattern of how your conviction shifts across days is the most useful thing about the data; letting you smooth that pattern after the fact would defeat the only reason it is being captured.
The other side of the constraint is that the prompt is small enough to fit any evening. Sixty seconds at the kitchen counter, ninety seconds in bed with a phone. The longest evenings · the ones where the question genuinely needs writing about · take as long as the writing takes, and that is fine. The shortest evenings ask for one tap and let you go back to your life. The product is not a journal you have to feed. It is a daily measurement of a position you already hold.
One tap a night for the duration you chose. The pattern of those taps, across the period, is the data the analysis reads on the final day.
A sentence, a paragraph, or nothing.
Underneath the four buttons is a note field, capped at three thousand characters. It is optional. Many evenings you will leave it blank. Many evenings you will write a sentence about what shifted that day. Occasionally you will write a paragraph because the question needed it. All of those are correct uses of the field.
On the final day, the analysis reads every note you wrote across the period and clusters the themes that recurred. Not the themes you predicted you would write about on day one · the themes that actually surfaced, in the words you used after work on a Wednesday or before bed on a Saturday. That clustering is what produces the synthesis paragraph: the one that names what you were actually arguing about underneath the surface question. The note field is the input to that synthesis. Sparse honest notes work better than verbose performative ones; if you write nothing, the trajectory of your votes still says plenty.
"Sunday morning at the park, watching a child running across the grass after a small dog. Caught myself smiling, then immediately panicked about money and time. Both feelings were real, in the same minute. Writing it down so it does not slip away by Tuesday."
The only thing visible across the table is whether you voted.
Through the period, the only thing either partner can see about the other is a single binary: voted tonight, or not yet voted tonight. The direction is invisible. The conviction band is invisible. The note text is invisible. The partner's name in the corner shows a soft "voted at 6:42pm" or a soft "pending tonight" and that is the whole channel.
This is deliberate. If you could see your partner's drift, you would anchor. The whole product breaks the moment that becomes possible. The voted/not-voted signal is the smallest piece of information that allows two people to feel they are running the decision together, without giving either of them anything they could use to second-guess their own vote.
What happens when you miss an evening.
You will miss evenings. You will travel; you will be sick; you will fall asleep with your phone face-down. A missed evening records as a skip and the analysis reads skips honestly · they are part of the trajectory, not a flaw in it. A decision with eight skips across thirty days tells you something about the question and something about the period you ran it through. We do not punish skips, and we do not paper over them.
If by the final week you feel the period has not been long enough, you may extend the duration once, by up to fifty per cent. You may not shorten. You may not vote retroactively on a day you missed. The integrity of the data is what makes the verdict on the final day worth opening; bending those rules to flatter a missed evening would make the entire product pointless.