Scenario: responding to a new regulation
A supervisory authority publishes a decision, it lands in your feed overnight, and you need to judge whether it touches your programme — and be able to show that you reviewed it the week it was published, not the week the auditor asked.
A directive amendment, a data-protection authority's decision, or an updated technical standard can unsettle a control mapping that took months to build. This scenario follows one such publication from the moment it appears in the feed to the point where a policy has been revised because of it. Two people do the work: Contributor Daniel, who watches the feed and prepares the ground, and Manager Ruth, who records the formal review decision. The journey crosses Regulatory changes, Compliance frameworks and Policies, and leaves a dated, attributed trail in each.
Settle one thing up front: Aegis pulls regulatory developments
automatically. A background worker polls the feeds registered
for your tenant on each feed's own cadence, writes every new publication in as a
regulatory change (a record of something an
authority has published), and scores it for relevance against the frameworks and
topics you track. Out of the box that means
CERT-EU Security Advisories,
EC Digital Strategy News and NIST Cybersecurity, plus
GDPRhub DPA Decisions where the licence allows it. An
Admin can register further RSS or
Atom feeds — another supervisory authority, say — in
Settings, giving each one an identifier, a name, a
URL, a region and a polling frequency. Nothing on this screen lets you type a
change in by hand. What needs a person is the judgement.
Before you start
Two gates stand in front of this screen, both carrying the same name. The
REGULATORY_INTELLIGENCE module has to be switched on for your
tenant, or the menu entry never appears; and your licence has to include the
matching feature, which ships from the Professional tier upwards. Without it the
page reads "Regulatory change tracking is not available on your current plan.
Upgrade to monitor and assess incoming regulatory changes." Above that gate, the
split is between reading, asking the AI, and deciding:
| Role | What they can do with a regulatory change |
|---|---|
| Viewer |
Read the feed, open any item's detail dialog, follow the source
link, and read any AI insights and action items already saved. A
Viewer cannot start an AI run, so the
Relevance Triage (AI) and
Suggested actions
buttons do not appear
|
| Contributor |
All of the above, plus run Relevance Triage (AI) and
Suggested actions
|
| Manager |
All of the above, plus Mark as Reviewed,
Analyze impact, Dismiss and
Restore to queue
|
| Admin | All of the above, plus registering the feeds that are polled and switching on the daily regulatory digest email in Settings |
The write actions sit behind the regulatory:write permission. Every
AI button additionally needs the ai:execute permission, because
each run spends one of your tenant's AI credits; Contributors and above hold it,
Viewers deliberately do not. Analyze impact needs both. If AI is
switched off for your tenant, Relevance Triage (AI) and
Suggested actions are hidden for every role. Where a role lacks a
permission, the button is not rendered at all — so Daniel opens the same dialog
Ruth does and sees fewer buttons on it. Anyone can keep watch; signing off that
a change has been reviewed rests with a named, accountable person.
What's on this screen
/regulatory-changes is a single feed of incoming items — one table,
no folders, no create button. In the left menu,
Regulatory Changes sits at the foot of the
Compliance group. Across the top runs the global bar: your
organisation's name, global search, the language picker, the light/dark toggle,
a question-mark icon that slides in the help panel, a book icon that opens this
User Guide, the notification bell with its unread count, and the signed-in
user's initials and name with Sign out beside it. A round purple
button at the bottom right opens the AI assistant chat; it is there on every
screen.
The page header reads Regulatory Changes over the line "Track
incoming regulatory updates and assess their impact on your compliance posture",
with a single action on the right: Relevance Triage (AI). Below it
sits the filter bar — a Search regulatory changes... box with its
own Search button, an All Sources dropdown built from
the sources actually present in your data plus the feeds in your registry, and a
View dropdown set to Active that you can switch to
Dismissed. The table carries five columns: Title,
Source, Relevance as a coloured percentage,
Published, and Status. There are no per-row buttons —
clicking anywhere on a row opens its detail dialog. The captured screen shows a
queue part-way through being worked: rows from several authorities — European
supervisors, the Belgian cybersecurity centre, data-protection decisions from
GDPRhub DPA Decisions, CERT-EU Security Advisories,
NIST Cybersecurity and EC Digital Strategy News —
published over roughly two months, newest first. Relevance badges appear in all
three colours, and the Status column mixes grey
Pending Review badges with green Reviewed ones. Note
that a high score does not mean an item has been dealt with: on the captured
screen the top row reads 95% and is still pending. Scroll past the table, below
the fold of this capture, and two more panels follow — the AI insights saved
from a triage run, and the action items raised against them.
Step 1 — Daniel opens the feed and works out what to read first
The change arrives on its own. Daniel's job is not to find it but to decide which items in the queue deserve his afternoon.
-
Open the module. Expand the
Compliancegroup in the left menu and selectRegulatory Changes. The feed loads twenty items to a page, newest first. -
Ask for a reading order. Select
Relevance Triage (AI)at the top right, thenRun Relevance Triagein the panel that opens. Aegis ranks your open — that is, non-dismissed — changes by stored relevance, review state and recency, then drafts a plan naming a next step for each: review it, assess its impact, or dismiss it. The ranking is computed, not written by the AI.Save as recordkeeps the plan in the saved-insights panel below the table. -
Narrow the feed. Type into
Search regulatory changes.... The table filters as you type — the match runs over both the title and the summary text — and theSearchbutton beside the box submits the same term. Or pick one authority fromAll Sources. Daniel filters to the issuing authority and finds the row he wants, stillPending Review. - Check the guide if a column is unfamiliar. The book icon in the top bar opens this User Guide inside Aegis, at the chapter for the screen you are on. The question-mark icon beside it slides in the shorter help panel instead.
-
Confirm whose account you are in. The name at the top right
is the account every action from here on is attributed to — and, for Daniel,
the cue that the write actions will not be available to him. The captured
screen is signed in as a Contributor, which is why
Relevance Triage (AI)is visible.
On arrival the percentage is computed by keyword match against the
frameworks your tenant tracks, your sectors, and how recently the item was
published: green below 40, amber from 40 to 69, red at 70 and above. Where
an impact analysis has since run, the figure you see is that analysis's own
overall relevance instead. The detail dialog says which it is: a small badge
beside the score reads Heuristic score,
AI score or Manual override, with the date it was
set. A later writer never replaces an earlier one of higher standing — a
manual override beats an AI score, which beats the heuristic — so a later
feed re-scan cannot quietly undo an analysis or an override. Either way it
helps you triage — read the red rows first — but it decides nothing. A low
score on an item that turns out to matter is exactly why a person still
reads each change.
Step 2 — Read the change and draft the actions it implies
Clicking a row opens a dialog headed Regulatory Change. Daniel
reads it in full before anyone records a decision.
-
Open the change. Click anywhere on the row. The dialog
opens with the relevance figure and its band across the top — for example
95% — High Relevance. -
Read the record.
Detailsholds theSource,Identifier,Publication DateandRelevance Score, the last with its provenance badge (Heuristic score,AI scoreorManual override).Summarycarries the publication's own text, tidied of feed markup;Categoriesshows the feed's own tags, plus the key of any framework a later impact analysis flagged; andSource LinkoffersView original document, which opens the authority's page in a new tab. Read the original — the summary is an extract, not the decision. AMetadatasection gives the record'sCreatedandLast Updatedtimes. -
Ask for a draft checklist. Select
Suggested actions. Aegis turns this change, plus any impact analysis already on record, into a prioritised list of concrete compliance actions with an owner hint for the top items. Saving it keeps the checklist on this change, in the insights and action items panels at the foot of the dialog. This needs read access plusai:execute, notregulatory:write, so Daniel can prepare the ground before handing over. A Viewer does not see this button. -
Hand it to a Manager. Daniel sees no
Mark as Reviewed,Analyze impactorDismissbutton — they are not rendered for his role. He confirms the change looks relevant and passes it to Ruth.
Step 3 — Ruth analyses the impact and records the review
Ruth opens the same item. As a Manager she holds
regulatory:write and ai:execute, so three further
buttons appear along the top of the dialog: Mark as Reviewed,
Analyze impact and Dismiss.
-
Look at what is already there. Where your licence includes
the AI assistant, newly ingested changes are queued for impact analysis on
arrival — twenty-five per scan run at most, so a large ingestion leaves the
remainder unanalysed. The
Impact Analysissection may therefore already be filled in. If it readsNo impact analysis yet, selectAnalyze impact. If you have not yet acknowledged your organisation's AI transparency notice, it opens first; once you accept it, the click goes through, and cancelling leaves the analysis un-run. Aegis then queues a background job and confirms "Impact analysis queued — results appear here when ready". -
Reopen the change once the worker has run. The section then
shows a written summary, an
Overall relevancefigure, when it wasAnalyzed, and a list ofAffected frameworks, each with its own percentage. -
Judge the reading. The analysis is a draft opinion. Ruth
weighs it against what she knows of the programme and decides whether she
agrees. Running it rewrites this change's own relevance score — unless a
Manual overrideis on record — and adds the flagged framework keys to its categories, so the row's percentage may move — but no control, policy or review state has changed. -
Mark it reviewed. Select
Mark as Reviewed. The button is replaced by a greenReviewedbadge, aReviewsection appears withReviewed AtandReviewed By, and the row'sStatusin the feed turns green. That dated, attributed record is the point of the whole step. -
Or set it aside honestly. If the change does not touch you,
select
Dismissinstead. The item leaves theActiveview and stays reachable underView→Dismissed, where its row carries an amberDismissedbadge; the dialog gains aDismissedsection withDismissed AtandDismissed By. The button asks no question first, so nothing records why — theReasonline appears only where one was supplied through the API.Restore to queuereverses it. Dismissing is a recorded decision, not a delete.
The AI assist
Three AI reads appear in this journey: the triage plan on the list, the suggested actions in the dialog, and the impact analysis. All three produce text for a person to weigh. None of them marks anything reviewed, dismisses an item, or edits a control, a policy or a framework mapping. The triage plan and the suggested actions write nothing at all until you save them; the impact analysis is the one exception to watch — it stores its findings on the change and, in doing so, overwrites that change's relevance score (unless a manual override is on record) and adds the flagged framework keys to its categories. All three reason only over your own records — they never fetch outside data and never invent a change that is not in your feed. Each run is metered against your tenant's AI credits; see the AI assistant chapter for how that budget works.
Step 4 — Carry the change into your frameworks and policies
Marking a change reviewed records that you saw it. Acting on it happens in the
modules it touches, starting from the Affected frameworks list.
- Open the affected framework and update its controls. Go to Compliance frameworks, open the framework the analysis flagged, and filter its controls by status, domain or owner group. Where the change tightens an expectation, update the control's implementation status and re-check its mappings — because one control can satisfy more than one framework, a single update may close a gap in several at once. See Control mapping for that view.
- Revise the policies. In Policies, edit each policy the change touches. Every revision runs through its own draft, review and approval lifecycle, so the update is itself approved by a person — another dated, attributed record.
- Refresh the evidence behind the affected controls. Where a control's proof no longer matches the new expectation, attach a current artefact in Evidence. An auditor will ask what changed as well as when you noticed.
- Close the loop on the action items. Work through the checklist saved on the change. The change record itself has no "addressed" status, so the action items are what tell you the response is finished.
Frameworks such as ISO 27001 (clause 6.1.3), NIS2 (a European cybersecurity
law) and DORA (a European financial-sector resilience regulation) expect a
documented way of monitoring and responding to regulatory change. An auditor
examines the trail as much as the source — the feed, the dated review, who
reviewed it, and what changed downstream. Marking each relevant change
Reviewed in the week you see it is what builds that trail; the
audit log keeps its own immutable copy.
Tips and limits
-
The review status is deliberately narrow —
Pending Review,ReviewedorDismissed. There is no "under assessment" state between them, so anything you want chased has to become an action item. -
Mark as Reviewedis one-way: there is no un-review. Leave an itemPending Reviewuntil someone has genuinely read it. - Nothing here tracks an effective date, and no reminder fires when one approaches. Record that date as a dated action item instead.
-
The impact analysis runs on a background worker, so it appears after a
delay. A failed parse shows as no analysis rather than an error — and,
unless a manual override is on record, it drops the change's relevance to 0
on its way out. If AI is switched off for your tenant, or your AI credits
are spent, the job is skipped without writing anything. Either way, an empty
Impact Analysissection is worth re-running rather than reading as "no impact". -
The dialog has no control for changing a relevance score by hand. A
Manual overridebadge appears only where someone withregulatory:writeset the score through the API; once set, neither the re-scan nor a later impact analysis replaces it. - Aegis raises a per-user alert only for newly ingested changes scoring 50% or above; below that, the item still appears in the feed but nobody is notified. Admins can also switch on a daily digest email in Settings.
- The feed reflects only the authorities your tenant polls, each on its own cadence. Never assume every change relevant to you will appear there — the registry is worth reviewing as your scope grows.
Where this connects
Each module in this journey has its own chapter: Regulatory changes for the feed and the detail dialog in depth, Regulatory alerts for the wider intelligence view, Compliance frameworks for the control mappings, Policies for the revisions, and Action items for the follow-up. When the auditor comes asking for that trail, Scenario: preparing for an external audit picks the story up. See also What each role can do and Settings, where the polled feeds and the digest are configured.