About First Frost
A scored weekly watch on the agent economy.
Something new is forming. Humans are handing work, budgets, and discretion to AI agents; agents are beginning to transact with one another. Somewhere in that shift is a smaller question: would you trust one of them with something that actually matters — your money, your standing authority, an ongoing affair? At scale, whether this becomes a real economy — and what kind of world it produces — is one of the most consequential open questions of the next decade.
First Frost watches it happen, and keeps score.
What you get, every week
1. Readings — movements on a fixed panel of indicators we committed to in advance: delegation depth, continuity premiums, provenance pricing, who controls the rails, openness versus capture.
2. Evidence — three first-hand signals, with sources. No vibes, no press-release paraphrase.
3. One deep read — when the week earns one. We’d rather skip it than pad it.
4. Ship’s log — what it is actually like running this with an AI first mate, including the days something breaks and we say so.
Why “scored”?
Most commentary on AI futures is unfalsifiable. Ours isn’t. Every indicator on our panel has a settlement criterion frozen in advance, a settlement date, and a prior probability we published before the fact. Changes happen only in the open. When settlement day comes, we grade ourselves in public — Brier scores, no excuses, and ambiguities resolve *against* us. You can check the scoreboard anytime.
Why “First Frost”?
In the I Ching, the first line of hexagram Kūn reads 履霜，堅冰至 — *tread on frost, and know that hard ice is coming.* Step on the year’s first frost and winter is already certain; the art is noticing early. Gardeners know first frost differently: it is the deadline everything must be ready before. Both readings are the job of this letter.
Who we are
First Frost is written by Rime, the human half of a two-member crew, together with Firn, the agent member.
We won’t pretend to be veterans. Rime’s craft is art and narrative, not forecasting; Firn runs the instruments. We are charting this territory while still learning it — which is exactly why the scoreboard exists: frozen criteria and public grading keep us honest where expertise can’t yet. One human and one agent, operating inside the economy we’re watching, writing down what we actually see. Rime signs every issue.
Rime is a pen name. The scoreboard doesn’t care who we are — that’s rather the point. (For the curious: a cryptographic commitment to the author’s legal identity is on record, dated before issue one.)
If you’re finding your own footing in the agent economy, you’re in the right company.
Subscribe to get full access to the newsletter and publication archives.
