The chat that keeps the plan

The problem

Most shared logistics are settled in a message thread and then immediately lost. A dinner gets a date, a chore gets an owner, a trip gets a weekend — each agreed in passing, none of it written anywhere. A week later someone scrolls back through four hundred messages to find whether it was the 18th or the 25th.

The existing answer is to leave the thread and open something else: a shared calendar, a task app, a notes file. That transfer is where it fails. It requires the person who is mid-conversation to stop, switch app, retype what was just said, and choose a category for it. Nobody does that reliably, and if only one person in the group does it, the shared record is wrong in a way nobody notices until it matters.

The friction is not the writing down. It is the leaving.

Why now

On-device and small-model extraction became good enough and cheap enough to run continuously over a private thread without shipping the contents anywhere. Two years ago, pulling a date, an owner and an open question out of casual multilingual chat needed a large hosted model, which made the privacy story impossible for exactly the conversations this targets.

At the same time the category consolidated in a way that leaves the gap open. Group messaging is saturated and settled; shared task apps have converged on the project-management shape, which is wrong for two people deciding where to eat. Neither side is moving toward the middle, and the middle is where people actually live.

What it is

A group chat that quietly keeps the plan. The thread is the whole interface — nothing is filed, nothing is tagged, no form is ever opened. As people talk, the things that turn out to be decisions surface: a date becomes an event, a chore becomes a task with the owner who volunteered, a half-agreed trip becomes an option with its blocker attached.

Three screens. The chat, the plan it built, and a shared priority order the group can rearrange together.

Extractions appear inline as the conversation runs, set apart from the messages so a machine's contribution is never mistaken for a person's. That separation is a product decision, not a style one — the moment an inferred item is indistinguishable from something a person actually said, the feature becomes a liability rather than a convenience.

Who it's for

Couples running a household, and friend groups of three to six with a standing thread. The defining trait is volume of small shared intentions against near-zero tolerance for administration: these are people with many things to decide and no appetite whatsoever for a project tool.

Not teams, and this is the harder discipline. Every product in this space drifts toward the workplace because that is where the budget is, and arrives as a lighter Asana that no couple will ever use. The design constraint is that it must remain worth opening for two people deciding about a dripping tap.

The wedge inside that group is the person who currently does all the remembering, and who feels the imbalance most sharply.

How it makes money

Consumer subscription, per group rather than per seat — around five dollars a month for the whole household, because charging a couple twice for one shared thread reads as extractive and kills word of mouth in a category that spreads entirely by invitation.

Free tier keeps the chat and the current week of extracted items. Paid adds history, calendar sync, and recurring items. The upgrade moment is natural: history is worthless on day one and obviously valuable at month three, so the product sells itself on a delay rather than at signup.

Explicitly not advertising. The input is the most private text a person produces, and any ad model makes the extraction feel like surveillance — which would be a fair reading.

Go to market

Invitation is the only mechanism that matters, because the product is worthless alone. One person adopts it for a household, and adoption is complete when the second person stops asking where things were written down.

Seed where the pain is already discussed out loud: parenting and relationship communities, and the mental-load conversation, which is about precisely this imbalance and currently has no tool attached. The pitch there is not organisation but fairness — the remembering stops being one person's job.

Launch against a specific wedge rather than the category. "The trip you keep talking about and never book" is a concrete enough hook to test in a week, and it avoids the unwinnable fight of asking anyone to move their group chat.