A competitor corpus that keeps itself current

Capture the category on a schedule so the reference set you query is this month's, not the one you built in March.

A corpus you built once silently expires. The value of asking it questions depends entirely on when it was last refreshed, and nobody remembers to refresh it by hand.

Why this will not look generic: The answers come from a reference set with a known capture date rather than from whatever you last happened to look at, which is the difference between a finding and a recollection.

What you walk away with

A reference set with a capture date you can trust
A running record of what changed between captures
Questions you can ask of the category at any point, not just when you last looked

Typical run: monthly or quarterly. Monthly is usually enough; most categories do not move fast enough for weekly to show anything but noise.

This one happens in one conversation

You bring: A corpus that has been refreshing on a schedule for a few months.

Open with: Compare this month's capture against the one from three months ago. What moved, and is it one brand changing or the category moving?

What the thread holds on to:

Both captures at once, which is what separates a category shift from one brand's redesign
What it already attributed to a single outlier
The running account across captures, which gets more useful every month

Then you can just ask:

Is anyone moving toward where we already are?
What appeared this quarter that was not here before?
Which of these changes should we be reacting to and which are noise?

What the reverse actually gives you

Point it at: This month's scheduled capture, read against the one from three months ago.

What comes back (representative example): Compared against the capture from three months ago, two things moved and one did not. The share of the set using a soft frontal key fell from sixteen of twenty to eleven, and every one of the five that changed moved the same way, to a hard side key with a visible shadow. Palette discipline tightened at the same time: the average colour count dropped from four to just under three. Composition did not move at all, with the subject still centred in nineteen of twenty. No single brand accounts for the shift on its own, which is what separates a category moving from one competitor rebranding.

Things you would never have thought to type:

soft frontal key falling from 16 of 20 to 11
all five that changed moving in the same direction
composition not moving at all while lighting did
no single brand accounting for the shift

What you change: Nothing yet. One capture is a snapshot; the diff between two is the part worth acting on.

How it works

Capture on a schedule (In your code) - A cron job against a fixed URL list. The list is the thing to maintain, not the captures.
Ask it what moved (In Claude) - The comparison between captures is the deliverable. A single capture is just a snapshot.

Cost and time

This workflow runs at 1 reverse + about 1 minute. Your result carries automatically from each step into the next, so there is no re-uploading between tools. Reverses are billed from your reverse meter and generation steps from your AI Credits, both bundled in every pay-as-you-go credit pack.

Browse all guided workflows or all individual tools. Pay as you go, no subscription, AI Credits never expire.