FIELD NOTE 02 / OCTOBER 31, 2026
A rising keyword isn’t
a business model.
Search interest helps us ask better questions. It does not tell us what people will buy. Here is the actual 30-day snapshot behind our first categories.
What we found: US interest in “AI agents” was 15.6% higher in the last seven days than in the first seven. Its latest week cooled. Worldwide, the same term fell 26.4%. None of the 25 researched series passed our sustained-growth screen.
One window. Two very different pictures.
October 1–30, 2026 · Web Search · All categories · Search terms, not topics. This is a dated research snapshot, not a live feed.
Data source: Google Trends. Normalized 0–100 interest within the five-term query group; not search volume. Region changes also change the normalization base.
The US core group
| Search term | 30-day mean | Last / first 7 days | Interpretation |
|---|---|---|---|
| AI tools | 74.83 | +7.9% | Higher relative interest, uneven |
| AI agents | 62.87 | +15.6% | Monthly gain, latest week cooling |
| AI video generator | 40.30 | −7.5% | Lower in this window |
| vibe coding | 14.73 | +32.1% | Lower base, latest week cooling |
| AI website builder | 4.63 | −28.9% | Lower in this comparison |
Commercial intent needs its own evidence.
“AI tools directory,” “submit AI tool,” “startup directory” and “promote startup” had many zero-index days in the commercial query group. Removing the larger “product launch” term did not remove that sparsity. “Product launch” itself fell 32.2%. We use directory phrases where they describe the product naturally; we do not treat them as a proven traffic engine.
How we calculated it
Growth = (mean of October 24–30 ÷ mean of October 1–7 − 1) × 100. A zero starting mean produces no percentage. Our own sustained-growth screen requires fewer than six zero days, at least 10% growth, a positive daily regression slope, no latest-week decline, and at least 60% positive changes in rolling seven-day averages. This is our research definition, not a Google classification.
Google Trends uses sampled, normalized relative interest. A 100 is the peak in a query group, not 100 searches. Repeated requests and changed comparison groups may produce different scales. Zero can mean insufficient data. Thirty days does not establish a durable market trend or advertiser willingness to pay.
Inspect the source
The daily files below were converted from the official Explore page’s accessible data tables, read on October 31, 2026. They are research exports, not files downloaded using Google’s export button. The original tables, settings, screenshots, all five groups, and calculation script are retained in our research archive.
- Open the official US query
- Open the official worldwide query
- US daily values (CSV) · Worldwide daily values (CSV)
- Google Trends data FAQ · Google’s export and attribution guidance
Paid placement is separate from this research. Advertisers cannot buy a better Google Trends score or an editorial recommendation here.