Every claim carries the page it came from.
Magdox does not summarise what it thinks it knows about a competitor. It reports what it observed, names where it saw it, and keeps the capture so the claim can still be checked months later.
We capture the page
On your plan's schedule, whether or not anything changed. The checks where nothing moved are kept too, because they are what makes a date range meaningful.
We hash what we captured
The stored file is named by the SHA-256 of its own contents, so a page edited later cannot be passed off as the one we read.
Every sentence cites a capture
A claim in a brief or a battlecard is linked to the observation behind it. One that cannot be linked is marked, not quietly shown.
You open the receipt
The source URL, the moment we observed it, the passage it was read from, and the hash. Exportable for anyone who has to check it.
What is actually stored
- The page as text, normalised so that whitespace and re-ordered markup do not read as a change.
- A screenshot, where the page renders. Some do not, and we would rather say so than show you a blank frame.
- A SHA-256 of the content, which is also the name the file is stored under, so the store checks itself on every read.
- The moment we looked - including the checks where nothing had changed, which are what make a date range mean anything.
What this does not claim
- It proves what a page said and when we read it. It does not prove why the company changed it, or what they meant by it.
- A sentence that cites a source can still be wrong in emphasis. Sourcing is a floor, not a guarantee, and we would rather say that here than have you discover it.
- We read publicly accessible pages only. Nothing gated, no accounts created, no sign-ins.
- A claim we cannot tie to an observation is shown as unsourced rather than quietly presented alongside the ones we can.
How long the record is kept
The window is the plan, and it is enforced rather than advertised. An organisation can shorten its own window in settings; it cannot extend it past what the plan includes.
- Watch
- 90 days of change history
- Track
- Two years of change history
- Panel
- Unlimited history retention
Why an assistant cannot do this
A general assistant can read a competitor's pricing page today and describe it well. It cannot tell you what that page said last month, because it was not watching, and nothing it does now can recover a page that has since been edited. That is not a gap that closes as models improve. It is the difference between reasoning and observation, and it is the only part of competitive intelligence that has to be accumulated rather than generated.
Which is why Magdox is built to be the thing your assistant asks, rather than a replacement for it. Connect one, and its answers about your competitors come back with the source and the date attached.
Start the record today.
The archive only helps if it was running before you needed it. Setup takes minutes and the price is on the pricing page.