Why this exists

The Fort Mill School Board’s May 18, 2026 meeting ran three hours and thirty-four minutes. Most people watched none of it. I’m guessing a lot of my neighbors didn’t either.

Almost every consequential decision in our daily lives happens in a room most of us will never sit in. The school board sets the calendar your kids live by, votes on millage that shows up on your tax bill, and approves the construction of buildings your family will spend years inside. Town council decides whether a developer can put 400 units up the road, what the police budget will be, and which streets get repaved this summer. These are not abstract policy questions. They are immediate and local. And the official record of how they were decided sits inside hours of video, hundreds of pages of agenda packets, and PDFs that arrive four to six weeks after the fact.

I built this site because I wanted to know what my school board had done, and discovered that the honest answer was that I would have to watch many hours of YouTube to find out. That is not a serious option for anyone with a job. It is also not a serious failure of government — the meetings are public, the agendas are posted, the videos are uploaded. It is a failure of legibility. There is a difference between a record being technically available and a record being readable, and we have built the first while skipping the second.

Every Monday morning at six o’clock, an automated pipeline checks for new meetings from each of the bodies tracked on this site. It pulls the official document or the meeting recording, sends it to a language model, and returns a short summary under a fixed editorial template: a paragraph of context, a bulleted list of every motion and vote, a list of items flagged for watching (subdivisions, rezonings, school facilities, policy changes), and the action items that came out of the meeting. Every summary links back to its primary source. If the model is wrong about a vote or a name, the source governs.

The watchlist categories aren’t neutral choices — they reflect what I personally want to know about local government. But the summaries themselves are. The model is instructed not to editorialize, not to predict, not to characterize anyone’s motives. If a board member said something inflammatory, the summary reports what they said and how the vote went. If a council adopted an ordinance, the summary records the tally and the dissent. There is no spin on offer. If you find any, report it and I will fix it.

This is not journalism. Real reporters call sources, attend in person, ask follow-up questions, and chase down the story behind the vote. Nothing here does that. A summary is what happened in the public meeting, not why, and the difference matters. If you want context, motive, and accountability, read the local newspaper — and pay for it.

This is also not a political project. I do not have a position on the proposed millage, the academic calendar, or the bond issuance. I have opinions about them as a resident, but they don’t go on the site. A civic record that wears a side loses the audience it is most useful to.

And this is not a replacement for the source documents. Auto-captions get names wrong. Language models occasionally hallucinate dollar figures. If a decision matters to you — legally, financially, or otherwise — click the source link. The summary is meant to get you to the right meeting, not stand in for it.

Two things converged that made this possible. First, local governments are finally posting their proceedings online by default — in standard formats, on standard platforms. Second, language models got good enough to read a long agenda packet and produce a faithful summary in under a minute for pennies. Neither of those was true five years ago. Both are true now. A weekly digest of every council, planning commission, zoning board, and school board in a county is suddenly possible for the cost of a cup of coffee a month.

That means it should exist. If it isn’t built by someone, the gap between what is technically public and what is practically readable stays open. I built this for Fort Mill and York County because that is where I live. The platform is meant to work anywhere. If you are reading this from another county and wishing your own school board were legible, write to me.

Three things.

Read with the source link open. Treat the summary as a way in, not a substitute. If something matters, verify.

Report errors. Speech recognition is imperfect. Language models are imperfect. Together they are more imperfect. If you spot a wrong name, a misstated motion, or a vote tally that doesn’t match the recording, write to corrections@localrecord.org with the meeting URL and what is wrong. Corrections post with a visible timestamp.

Tell a neighbor. This site is free, has no advertising, and isn’t collecting your data. The only reason for it to exist is for residents to use it. If you find a summary that surprised you in a useful way, send it to someone.

— The editor, Fort Mill, SC
May 2026