The meeting summary is for the people who weren't there

Most meeting notes aren't the problem. People write things down; the raw material almost always survives. The real problem is what gets lost between the conversation and the work — and one specific reason it gets lost is worth naming.

Meeting summaries are written by the person who ran the meeting, from the perspective of someone who was in the room. And the person who was in the room is precisely the person who will never need the summary.

They already know what happened. Their notes are reminders — compressed, allusive, full of context that lives in their head. priya — faq by thu?? is a perfectly good note for the person who typed it mid-conversation. For everyone else, it's a puzzle.

The summary's real readers are absent by definition: the teammate who skipped this one, the stakeholder who only wants the outcome, the new hire reading up three months from now. Write for them and the summary changes shape entirely.

A reminder and a record are different documents

Here's the pushback: doesn't the organiser need the record too — for accountability, for the follow-up?

They need parts of it. They need the action list, because they'll chase it. They need the decisions, because they'll be asked. But those are the structured, assignable parts of the record — not the explanatory summary wrapped around them. The summary as prose exists to transfer understanding, and the organiser has nothing to transfer to themselves.

That distinction explains a familiar failure. Notes written as reminders work for exactly one reader for about one week. Then the context they leaned on fades — even for the author — and the team is left with a document that says budget → finance, not us and no way to reconstruct why. The note didn't fail. It was never written for the reader who eventually showed up.

What an absent reader needs, in their order

Someone who missed the meeting asks four questions, in a fixed order:

• The context — why was this discussion happening at all?
• The topics actually covered — what ground did it walk?
• The decisions that were made — what's now settled?
• The next steps — who owes what, by when?

That order matters. A decision without its context invites relitigating; next steps before the decision read as arbitrary chores. The absent reader needs the story in one minute, front to back — not forty accurate minutes of transcript in the order things happened to be said.

This is the structure Scripta builds from what your team already has. Paste the transcript, or the rough notes typed at speed during the call, and it produces the four answers — no rewriting, no formatting, no copying tasks into a second tool while the meeting is still fresh enough to remember.

The input can be messy, because meetings are messy — and because the author shouldn't have to spend twenty minutes translating reminder-notes into reader-prose. That admin gap is exactly where the translation usually dies.

One thing the structuring never does is upgrade the truth. A question the room parked stays parked in the summary; proposed decisions carry the line they came from, and a person confirms them before anyone treats them as the record — the AI shows its work rather than smoothing ambiguity into confidence the meeting never had.

Write for the reader who arrives in three months

The furthest-away reader is the strictest test. A new teammate reading the record in November has no shared context at all: every pronoun, every "as discussed", every unstated why is a dead end for them.

You don't have to write for November by hand. You have to keep the record in a shape where November can be served: context attached, decisions with their reasoning, commitments with owners and dates, and the discussion that produced them one click away rather than in someone's memory.

Because meetings shouldn't end with more admin work. They should just end — and the record should do the remembering, for whoever opens it next.