This week in trends · Aug 6 – 15, 2026 · 6 min read

What's trending now — and how last week's list actually did

What actually gained speed between Aug 6 and Aug 15 — and, for the first time, a scorecard on the five signals we published last week, including the three that went against us.

Published Aug 15, 2026 · by the Lolik team, from our own radar data

Key takeaways

  • We graded our own calls. Of the five signals we published on Aug 5, two kept climbing, one faded from its peak, one never confirmed, and one fell out of tracking entirely.
  • HeyGen keeps compounding — 15,978 views/day (Jul 24) → 23,843 (Aug 5) → 39,034 today, still rising after 30 tracked days.
  • The pool stayed cold for a third window: 4 of 942 tracked signals held the breakout label, and 854 are decaying.
  • Watch-phase signals do graduate: Hormozi moved from last week's watchlist into breakout at 10,407 views/day — slowly, without a spike.
  • All figures measured Aug 15, 2026 on our own radar — units and tracking windows stated inline.

942

signals live-tracked in this window

239

new signals first seen in this window

4

in breakout phase

6

niches covered

First: how last week's list actually did

Every weekly edition here prints numbers that can be checked later, so this one starts by checking the last one. Below are the five signals we published on Aug 5, with the figure we printed then and what the same signal measures today. Two held up. Three did not — and those are the rows worth your attention, because they show how quickly a "hot" window can close.

Graham Stephan — "The Best Free Money Hack"

Decaying

57,345 views/day (Aug 5, Breakout #1) 21,065 views/day (Aug 15)

We published this one within a day of its peak — it measured 56,396 views/day on Aug 5 and has fallen every single day since. Our own rule says a breakout is a phase, not a promise, and this is what the bad side of that looks like: by the time a finance video is this loud, the window is closing, not opening.

HeyGen (AI video)

Breakout

23,843 views/day (Aug 5, Breakout #2) 39,034 views/day (Aug 15)

Third edition in a row on this list, and it has not stopped: 15,978 views/day on Jul 24, 23,843 on Aug 5, 39,034 today, still rising after 30 tracked days. AI-avatar video is the most durable wave on our radar.

Alex Hormozi — "What Working Hard Actually Looks Like"

Breakout

8,872 views/day (Aug 5, Watch #2) 10,407 views/day (Aug 15)

The one that graduated. It sat on the watchlist for weeks as a flat burn, then confirmed — a slow, boring climb into breakout. Not every confirmation is a spike.

Jeremy Ethier — fitness

Decaying

92,746 views/day (Aug 5, Watch #1, biggest raw number) 98,732 views/day (Aug 15)

We said it would either graduate or fade, and that the watch label was the honest call. It peaked at 101,902 views/day on Aug 13 — a 10% gain on ten days — then rolled over without ever confirming. Still the biggest raw number we have ever tracked, and still never a breakout.

Agent-Reach

Dropped out

411 stars/day (Aug 5, Breakout #3) no longer tracked (Aug 15)

It stopped appearing in our ingestion and fell out of the 30-day retention window, so we have no honest reading for it today — not a decay we measured, an observation we lost. Flagging it rather than quietly dropping the row.

Both columns measured on our own radar — the first on the day we published it, the second on Aug 15, 2026.

Where the 942 live signals stand

Decaying 854Watch 84Breakout 4

Most of what spikes never confirms — 4 of 942 signals held the breakout label in this window and 854 are decaying. Counts measured Aug 15, 2026.

The confirmed top 3 — signals with sustained, observed velocity

Three signals cleared the confirmation bar in this window: velocity sustained across enough tracked days to be marked observed, not just spiking. A fourth signal holds the breakout label but is not marked observed, so it stays out of the ranked list rather than padding it. Ordered by overall score (velocity × acceleration × durability), not raw numbers alone — and the bar under each velocity is scaled within its own unit group, because stars/day and views/day never share a scale.

#1

HeyGen (AI video)

BreakoutAI
39,034 views/dayYouTube · tracked 30 days · our full retention window
The most durable signal we have: a month of tracking, still accelerating, and up 2.4× since we first published it on Jul 24. When a wave holds this long, the opportunity is no longer the news — it is the second-order content (comparisons, workflows, "I used it for 30 days") that arrives while everyone else is still reacting to the announcement.
Watch on YouTube: HeyGen (AI video)
#2

Alex Hormozi — "What Working Hard Actually Looks Like"

BreakoutBusiness
10,407 views/dayYouTube · tracked 14 days · first seen Aug 2
Promoted off last week's watchlist after two weeks of steady climbing — no spike, no viral moment, just a floor that kept rising. Work-ethic content remains the most reliable base in the business niche: it does not explode, but it almost never dies.
Watch on YouTube: Alex Hormozi — "What Working Hard Actually Looks Like"
#3

firecrawl

BreakoutAI
197 stars/dayGitHub · tracked 12 days · first seen Aug 4
The web-scraping API for AI agents is the only developer signal that confirmed this window, and it is the third edition in a row it appears in. Agent infrastructure keeps being the quiet, durable pocket of AI demand — slower than the model-release news cycle, and far longer-lived.
Full tracked history

The watchlist — flagged, still confirming

Both signals below are flagged but unconfirmed, and both are developer tools sitting just under the bar for weeks. This is the phase where the timing edge actually lives, and also the phase with the most false starts — last week's biggest watch signal peaked and rolled over without ever confirming, while Hormozi quietly graduated. How to read this phase: our guide to spotting trends early.

#1

dify

WatchAI
125 stars/dayGitHub · tracked 26 days · still confirming
Agentic workflows and RAG pipelines in one platform — nearly four weeks of elevated star velocity without clearing the confirmation bar. The same shape Hormozi had before he graduated: flat, patient, not dead.
Full tracked history
#2

prompts.chat

WatchAI
124 stars/dayGitHub · tracked 30 days · still confirming
The prompt-sharing library (formerly Awesome ChatGPT Prompts) has held a steady star rate for our whole retention window without accelerating. A floor, not a wave — useful to know before you build a video around it.

What we'd do with this, practically

The clearest lesson of this window is a timing one: the finance breakout we published was already one day past its peak, while the AI-video wave we have carried for three editions is still accelerating after a month. If you are choosing where to spend a week of production, durability beats size — a signal with 30 days of sustained velocity gives you room to publish something considered, while a 57,000 views/day spike demands you already be filming.

For the AI and developer audience, agent infrastructure remains the confirmed pocket: firecrawl confirmed, dify and prompts.chat holding just underneath. For the full pipeline from a signal like these to an actual brief — finding topics before they peak and monetizing them realistically.

Measured, in public

In our own backtest, early-phase alerts kept growing over the next 5 days in 47.7% of cases vs a 2.2% baseline across every tracked signal-day — 21.4× more likely to keep growing than a random tracked trend (n=65 from 30 unique signals, Aug 6 – 12, 2026, a window with zero overlap with anything we published before). That sample is small — the conservative read is the bottom of the 95% interval, still . Across the wider Jul 2 – Aug 12, 2026 window (n=374) the same measurement gives 11.1×.

Point-in-time figures, re-measured weekly — full methodology and every call, wins and misses, in the public log.

Common questions

Why publish that three of last week's five signals went against you?

Because a feed that only shows its wins is marketing, not measurement. We published Graham Stephan at 57,345 views/day one day after its peak, and it is at 21,065 views/day now. That is useful information about how fast a finance window closes — and you cannot trust the calls that worked if we hide the ones that did not.

Only 3 confirmed breakouts again — is the radar broken?

No, the pool is genuinely quiet: 942 live signals, 854 of them decaying, and only 4 holding the breakout label (3 of which are marked observed and therefore eligible for this list). That is the third quiet window in a row, versus 44 of 898 in our Jul 18–24 edition. We publish the count we measure.

Why did the biggest number on the radar never become a breakout?

Jeremy Ethier's fitness video peaked at 101,902 views/day on Aug 13 and rolled over without confirming. Breakout in our system means sustained, accelerating velocity across enough tracked days — not size. A signal can be the largest thing we track and still never qualify, which is exactly why we label the watch phase honestly.

Can I see these signals live instead of waiting for the post?

Yes — the radar updates daily and the free tools run on the same data, no signup. The weekly post is the measured summary; the product is the live feed. Timing edge is a possibility, not a guarantee — outcomes depend on your execution.

Want the next list while it's still forming?

The free tools run on the same live data as the paid radar — no signup, no card. Start there; upgrade only if the timing edge earns it.