The evidence layer behind Compass.
See what Compass reads, how articles become measurements, how fresh the corpus is, and where the limits are. The methodology is public because the output is meant to be cited.
A selected, source-linked technology press corpus.
Compass does not measure the whole web. It measures a selected, dated corpus of the technology press — established outlets, specialist publications, and named primary sources. That boundary is the instrument. Measuring everything means measuring the noise too, and a signal drawn from an unbounded pile can't be compared to itself over time. A fixed, known corpus can — so this month's number and last year's number mean the same thing.
Compass measures a selected, source-linked technology press corpus — established outlets covering the field — alongside Enginerds' own weekly analysis. Every source is curated and linked, so each measurement traces straight back to the coverage behind it.
The frame is broad by design, so what it measures stays representative — from AI and semiconductors to cybersecurity and cloud infrastructure, and the business, funding and policy moves underneath them, in the US and beyond. Because every article is dated, movement is measured over months, not read off a single day.
Two kinds of source make up that frame, and they never blur. Most of it is independent press — and that's what shapes how Compass reads the framing of a story. A smaller, clearly-labelled set of primary sources, like the AI labs' and platform vendors' own posts, grounds first-hand launches and results, but stays out of that framing read. Wherever a signal rests on one, you'll see it in the source trail.
How the corpus is chosen. A source qualifies on a consistent bar — an established technology outlet, a specialist trade publication, or a named primary source — not on how loud any one story is. The set is reviewed on a set cadence: sources that meet the bar are added, and ones that go dark are dropped. The point isn't to be exhaustive; it's to measure the same defined body of coverage the same way every month, so comparisons over time hold.
From articles to signals, in four passes
Compass doesn't ask a model "what's important." It runs a deterministic pipeline that turns the article record into structured measurements — the same way every time, so the output is reproducible and auditable.
- 1
Enrichment. Each article is normalized and tagged to its topic threads and timeframe — the scaffolding everything else hangs on.
- 2
The players, identified. The topics and actors in play — companies, technologies, products, and the people behind them — are surfaced and tied back to the articles that name them.
- 3
Canonicalization. Aliases are merged — so "GPT‑4" and "GPT‑4o" resolve to one entity — and citations are linked into a source graph.
- 4
Signals. From that structure Compass computes the measurements you see — monthly presence and velocity, persistence, emergence, and the connectedness score that ranks how strongly each topic links to others. All arithmetic over the record; no opinions injected.
Current, dated, and preserved.
As new coverage lands, the measurements move with it — your dashboards always reflect the latest state of play. Nothing is thrown away: the full history stays intact, so you see not just where things stand, but how they got there. And every screen is stamped with the moment it reflects, so you always know how current your view is.
Every signal shows its evidence.
Compass never asks you to take a number on faith. Alongside every measurement, you see how much of the record stands behind it — so a broad, well-supported signal is never mistaken for an early, emerging one. That's what makes the output citable: you're always reading evidence, with its weight in plain view.
And Compass is careful about what a connection means. It tells you two topics show up together in the record — a real signal in its own right — while leaving the why for you to draw.
That discipline is the whole point. Compass measures a real, dated, source-linked body of evidence — and stops at the measurement. The conclusion is yours to draw.