Threads
- Momentum: gaining When an agent is the first reader of your docs
What does a product surface owe a reader that is software — and who controls the rail it arrives on?
Machine-mediated discovery moved from reading to selecting to transacting, and the front door moved with it — from published files to callable entry points, then past the account itself: Neon lets an agent provision a capped, 72-hour Postgres project with no human signup and the human claims it later. The measurement half arrived in early September (Val Town's first self-reported attributed AI-referral datum, Profound's 12% pricing-page citation share), and by the third week "does an agent pick you" had a vendor selling the answer — Armature's homepage now says "Get discovered" — while the transacting end reached consumer scale through Stripe's Link wallet inside Meta's Muse.
A second vendor reporting attributed AI-referral signups, or Val Town publishing its underlying numbers. The first datum landed 2026-09-02 — self-reported, partial, a share rather than a conversion rate — so the standard now needs a second instance or a denominator. · by end of Q3 2026
- Momentum: gaining How AI features get priced — seat, meter, data, and now the outcome
Is seat-plus-meter still the devtools default, is training data becoming a third thing developers pay with — and is the meter's unit moving from the token to the outcome?
Per-seat licence plus metered consumption settled in as the devtools default for AI features. Meta priced a discount as a data trade and no incumbent answered it by the August 31 clock, so the data axis reads as one vendor's beta mechanism. The meter widened in September (Atlassian metering automation steps, cap shipped a quarter before the bill) and then its unit moved: Sourcegraph bills Agentic Batch Changes per merged changeset and nothing for a rejected diff, while the token it replaces fell 23.2% in a month on Vercel's gateway and moves 71% by harness alone in UC Berkeley's HarnessTax study. The catch is that Sourcegraph's rate is not public. Two live tests: whether that number appears on the pricing page, and whether Copilot's prepaid seats from October 1 produce the first bill-shock story.
Does a *non-generative* feature get metered? Answered once on 2026-09-01 — Atlassian meters automation steps at $0.50 per 1,000 above a per-seat allowance, billing from December 3 — so the meter is a billing habit for at least one vendor. A second vendor metering a non-AI unit makes it the pattern; none by the end of the year leaves Atlassian as one company's billing reform. · by end of 2026
- Momentum: steady Measuring DevRel by influence, not attribution
If DevRel's effect cannot be attributed to a click, what four numbers should a program report instead — and is anyone reporting them?
The consensus moved to influenced pipeline plus activation, and the seller side started shipping the instrumentation: a Copilot dashboard with a named "Passive" seat segment, docs audits scored against the support queue, and AI-answer presence arriving as a discovery metric. The gap is that surveys still say most programmes cannot demonstrate impact.
A second AI devtool shipping passive-seat reporting, or a public renewal story that cites one. GitHub's dashboard is currently the only vendor putting an uncomfortable seat number in front of the buyer. · by end of Q3 2026
- Momentum: gaining Proof over adjectives — publishing the uncomfortable number
Does publishing the number that cuts against you actually outperform claiming the one that flatters you?
Developers fact-check in public, so the durable move is shipping something checkable — a dogfooding migration with its costs named, a benchmark that reports where you lose, a postmortem that prints the numbers that hurt. In late August the thread gained both halves at once: buyer research (LeadDev) saying unfalsifiable AI claims measurably repel skeptical engineering buyers, and vendors shipping proof as the campaign itself — a $1M attack-me bounty, a named security framework argued across a week of dated posts.
A second limits-raised case study — a vendor that dogfooded, hit its own published ceiling, and said so with the number.