A record is only true as of the day it was written.
Published regulatory material is watched for change. What reaches your record has been read by a person first, and carries the link it was found at.
The failure is quiet, which is what makes it expensive
Nobody misses a regulatory change on purpose. They miss it because nothing happened, and nothing happening looks exactly like nothing being wrong.
Guidance gets revised in place at the same address, so the link in your file still works and still looks current. A consultation closes without anyone circulating it. An expectation shifts in a way that changes what your evidence needed to say, and the evidence sits there reading perfectly well against a rule that has moved.
You find out when someone asks. By then the gap has a date on it, and the date is earlier than you would like.
What reaches you, and what has to happen first
The value is not in noticing. It is in what is done between noticing and telling you.
- ✓A person approves anything presented as factChange is detected automatically and recorded as a proposal. It becomes part of the record only after a human has read it and agreed. Nothing automatically concluded is ever shown to your firm as something a regulator said.
- ✓Every event carries the link it was found atTaken from the page actually read, never assembled or guessed. There is no route through the system that records an event it cannot point at, so anything you are told can be checked at source before you act on it.
- ✓Revisions in place are caughtGuidance replaced at the same address is the hardest kind of change to notice, because nothing about the link or the title changes. The content itself is fingerprinted, so a silent revision is visible as a revision.
- ✓Your evidence is flagged, not editedWhen something changes that bears on a principle you have evidenced, the affected evidence is marked for review. It is not rewritten, corrected or re-filed. The writing stays yours and so does the decision.
Silence is never reported as calm
A quiet week and a broken watcher look identical from the outside. That is the failure this is built around.
A monitor that reports nothing because it found nothing, and a monitor that reports nothing because it stopped working, produce the same empty inbox. The second is how a missed change actually happens, and it can continue for months while every report looks reassuring.
So the system is built to fail loudly. A source that answers but yields less than it should is treated as an error rather than a quiet day, and a run that finds nothing because everything failed is never allowed to look like a run that found nothing because nothing happened.
What it will not do
Most of the design here is restraint, and the restraint is the product.
It does not tell you what to do about a change. No recommended actions, no deadlines of its own invention. It tells you what changed, where it changed, and which of your principles it bears on.
It does not publish anything on your behalf. Nothing is sent to a regulator, and nothing about your firm leaves the record.
It does not claim completeness. No monitoring can promise it has seen everything, and a product that implied otherwise would be inviting exactly the reliance it could not support.