AI agents visited Vichy Laboratoires's pages more than any other brand's this week

What stood out this week

Up to three findings, chosen from a fixed list of candidates. A finding only appears when it clears a minimum count, so a large percentage built on a handful of visits is left out.

1. User-directed activity, as Cloudflare labels it, is 3.3× more concentrated in this panel than across Cloudflare's network
   Cloudflare labels 15.3% of the AI-agent visits to this site as "User Action", against 4.7% across its whole network. Both figures use Cloudflare's classification, so the two sides are comparable.
   Why this matters. It shows whether this site sees an unusual amount of the traffic Cloudflare attributes to a person's action, which affects how far these numbers can be generalised.
   What this does not show. This uses Cloudflare's "User Action" label on both sides, which is not the same thing as Optimly's user-directed group and will not match that number. Our side is out of visits to optimly.ai from the agents Cloudflare counts as AI agents (search-engine crawling excluded, to match the network figure); the network side is out of AI-agent visits across Cloudflare's whole network, as published by Cloudflare Radar.
2. "Someone using an AI product" moved down from 38.8% to 12.2% of all visits
   This is one of Optimly's five groups, measured against every visit Cloudflare flagged as automated on optimly.ai during the week. Visits from agents the company that runs them documents as firing when a person asks for something. This is the only group we connect to a person.
   Why this matters. A change in the mix means the reason agents are visiting has changed, even when the total looks flat.
   What this does not show. The previous week is a shorter sample scaled up to a seven-day equivalent, so small share changes are within the noise.
3. Telecommunications pages drew 5.6× fewer visits than the week before
   75 visits this week against 420 visits the week before, across every brand page we place in Telecommunications.
   Why this matters. A category moving on both a real visit count and a real change is the earliest signal we get that attention is shifting.
   What this does not show. The previous week is a shorter sample scaled up to a seven-day equivalent, so treat the size of the change as directional.

Optimly's five groups

Every automated visit lands in exactly one of these five groups, so they add up to the week's total. Share is a percentage of every visit Cloudflare flagged as automated on optimly.ai during the week — 146,188 visits this week.

Cloudflare's own labels for the reason for the visit

This is a second, separate grouping. Cloudflare publishes its own label for each AI agent it recognises, and applies the same labels across its whole network. These numbers are not a breakdown of the five groups above and will not match them, because they count a different set of visits and use a different denominator.

On this site

Share is a percentage of visits to optimly.ai from the agents Cloudflare counts as AI agents (search-engine crawling excluded, to match the network figure) — 118,593 visits this week.

Across Cloudflare's network

Share is a percentage of AI-agent visits across Cloudflare's whole network, as published by Cloudflare Radar. Pulled 2026-08-09.

Under Cloudflare's own “User Action” label, 15.3% of the AI-agent visits to this site carry that label, against 4.7% across Cloudflare's network. That is 3.3× more concentrated in this panel than across Cloudflare's network. Both sides of that comparison use Cloudflare's label, not Optimly's groups.

Which of our pages the visits went to

We could match 75,985 visits to a page we publish, 52.0% of all automated visits. The percentages in this table are of matched visits only, not of the week's total.

Which pages agents requested

Every number below counts page visits we observed in our own server logs: an automated agent asked optimly.ai for a specific page. A page visit records the request and nothing more. It does not show that any AI product mentioned, cited, recommended, or sold anything, and it does not show that a person asked a question.

• Topic pages requested: Agents requested 64 pages built around a buying topic In the week of Week of August 2, 2026, automated agents requested 202 buyer-question pages across 64 distinct buyer questions — 76% versus the prior week.
• Brand pages requested: Agents requested 17,541 pages for individual brands In the week of Week of August 2, 2026, automated agents requested 58,920 brand pages across 17,541 distinct brands — -4% versus the prior week.
• Brand data files requested: Agents requested the machine-readable file for 1,069 brands In the week of Week of August 2, 2026, automated agents requested 2,403 brand record pages across 1,069 distinct brands — 97% versus the prior week.
• Change reports requested: Agents requested week-on-week change reports for 1,054 brands In the week of Week of August 2, 2026, automated agents requested 2,372 brand delta report pages across 1,054 distinct brands — 139% versus the prior week.
• Category list pages requested: Agents requested 4 pages listing our categories In the week of Week of August 2, 2026, automated agents requested 98 category pages across 4 distinct categories — -44% versus the prior week.

Topics of the pages agents requested most often

These are short topic labels we assign to our own pages, so we can group page visits that keep coming back week after week. The label is our grouping of the page. It is not a person's exact question, and we do not know what anyone typed. Counts run from the week we started grouping pages this way, so the list builds up over time rather than reshuffling every week.

Solution cluster

• Cloud — 10 total (2 this week), first seen 2026-W30
• Ppc — 2 total (2 this week), first seen 2026-W32

Topic pages agents requested most this week

Each row is one of our own topic pages that an automated agent requested in the week of Week of August 2, 2026, ordered by how many times it was requested. The heading is the topic we assigned to the page. We do not know what any person typed, and a visit does not show that a person asked anything.

Visiting a Physical Casino or Racetrack to Place Bets in Person

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Using Personal Checking and Savings Accounts for Business Transactions Common in Sole Proprietorships

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Using Local Git Repositories and Manually Sharing Code Via Zip Files or Shared Network Drives

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Using Excel or Google Sheets to Manually Calculate Withholdings Net Pay and Tax Liabilities

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Hiring Specialized Machine Learning Consultancies to Build and Deploy Custom Models

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Managing Cap Tables and Stock Options Using Excel or Google Sheets

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Hiring Specialized AI Consultancies to Build and Maintain Bespoke Conversational Interfaces

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Hiring Human Transcriptionists or Using Internal Staff to Manually Type Out Audio Content

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Hiring An Independent Interior Designer to Source Custom Pieces From Trade Only Showrooms

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Managing Kubernetes Control Planes and Nodes Manually On Ec2 Instances or On Premises Servers

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Transcribing Audio Manually or Using Internal Staff to Type Out Recordings

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Managing Customer Health Scores and Renewals Using Excel or Google Sheets

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Accepting the Risk of Identity Theft Without Proactive Monitoring or Insurance

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Connecting a Laptop or Pc Directly to a Tv Via Hdmi Cable to Stream Content

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Using Personal Credit Cards for Business Expenses and Manually Reconciling Receipts

Requested 4 times this week, most often by AI Assistant (behavioral). Group: problem. Not seen in the prior week.

Since we started measuring

This index started grouping pages by topic in 2026-W28. This week (Week of August 2, 2026) marks 5 archived weeks of continuous measurement. All counts labeled "total" or "lifetime" on this page are rolled up across every archived weekly report since inception; counts labeled "this week" are the week of Week of August 2, 2026 only.

Brands whose pages were requested far more this week

No brands broke out yet. On a fresh install the WoW comparison window fills in after seven days; expect the first real spikes on the following Monday.

Which agents requested pages from Optimly this week

146,188 visits from automated agents this week, a change of 41% against the prior week.

Agents that fetch a page while a request is in flight

Agents that collect page text in bulk

These agents collect page text to build a model rather than to answer a specific request, so no person is waiting on them. Listed for completeness, kept separate from the numbers above.