Noise settles quietly into a compliance team's day. Articles about the same event fill the list row after row. After a bulk action the selection resets and the same records have to be ticked again. In a screening where dozens of profiles match, it gets hard to tell which note belongs to which profile. And notification emails keep arriving all day, many of them unrelated to the task at hand.
We built the September release to reduce that noise. The highlights:
- Articles about the same event are gathered in one row, with Exact and Broad matching and negative keywords.
- A summary that is enough to decide can be seen without leaving the case list, and bulk actions keep your selection.
- Notes can be written directly on a profile.
- Notifications are set per event.
- Rules can depend on each other and keep a version history.
- AI reaches Widget Builder, Network Analysis, profile-level agents and AI Suggest.
Adverse media screening
Grouping and match types
The problem. Because the same event was published again and again by different outlets, result lists were far more crowded than they needed to be, and namesakes slipped in.
What changed. News is grouped by event, and the team controls how broad each search is.
How does news grouping work?
When the same event is reported by many outlets, the articles about it are gathered in a single row. The row shows how many sources reported the event. Open it and the articles are listed one by one, each with its source, category and date. Articles are not removed, only grouped, so every source stays on the record.
Grouping is an organization setting, Group Similar News, so the team decides whether results are grouped.
What do Exact and Broad mean?
- Exact: the full name appears together in the article. Word order doesn't matter.
- Broad: all parts of the name appear somewhere in the article, in any order or position.
The match type applies to the name only, not to the category keywords. An administrator sets a default for the organization, and an officer can change it for a single scan.
How do negative keywords work?
Articles that contain any of your negative keywords are left out of the results, even if every other condition is met. A common use is a namesake: an athlete, a brand or a place with the same name as your customer. The list applies to every scan, to Load More and to ongoing monitoring runs on existing cases.
The scan detail tells you how many retrieved articles were left out by the match type or the negative keywords. If that number looks high, you can widen the match type and see more.
Why it matters. The list grows with the number of events, not the number of articles, and every source stays on the record.
Case management
Inline summary and bulk actions that keep your selection
The problem. To see a case summary you had to open the case, and after a bulk action you had to select the same records again.
What changed. Case lists now open inline and keep your selection.
What does the inline summary show?
Each row in a case list now has an expand arrow. Open it and you see a summary that is enough to decide simple cases without opening them:
- The number of sanctions, PEP, watchlist and adverse media matches.
- The number of exact matches.
- The scan details.
- The five profiles with the highest match rate.
The summary works across Name Screening, Advanced Adverse Media, Transaction and Fraud Monitoring, Ongoing Monitoring, Swift and Crypto Address cases, on My Assigned Cases, Alert and Case Management and Reports, and on Live Transactions.
How do bulk actions keep your selection?
Until now, the selection reset after every bulk action, and the same records had to be ticked again. Now the table doesn't reload. Each change is held as pending, and a bar at the bottom of the page shows how many changes are waiting, with the old and the new value. You can undo a single change, cancel them all, or save them all in one go. If you try to leave the page with unsaved changes, you get a warning, and actions that will go for approval are marked as such.
Why it matters. Simple cases are closed from the list, and the analyst has more time for the hard ones.
Profile notes
The problem. Notes were kept only at case level, so in screenings with many profiles it got hard to tell which note belonged to which profile.
How do profile notes work?
In a screening with many matching profiles, a note at case level doesn't say which profile it is about. You can now write a note directly on a profile from its detail view, in Name Screening and Ongoing Monitoring. A profile can carry several notes, and each one is written to the activity log. The Memo tab on the right shows case notes and profile notes under separate headings. Generate AI Note works with the profile's context, so a drafted note uses the data of the selected profile.
Why it matters. The reasoning about a profile stays next to that profile and can be traced in an audit.
Notifications and rules
Per-event notifications
The problem. Notifications had a single switch, and turning it off stopped every notification email.
How do per-event notifications work?
The Notification Preferences card on My Profile lists the events you can be notified about: case assignment, a mention in a note, a shared report, ongoing monitoring, a transaction alert, an approval request, a completed batch and Agentic AI. For each event you turn in-app and email notifications on or off separately. Two master switches sit on top. Turning one off doesn't delete your choices, so they come back when you switch it on again. Account security and export delivery notifications stay on for security reasons.
Why it matters. Each person gets only the notifications that help their work.
Rule chaining and version history
The problem. Scenarios where a rule should run only when another rule fires couldn't be built, and the history of changes wasn't kept.
How does rule chaining work?
You can make a rule depend on another rule. If the rule it depends on doesn't fire for a transaction, the dependent rule is skipped. Dependencies are drawn on a canvas: you connect rules visually and save the map. This makes multi-step scenarios possible without squeezing them into one large rule, for example "check the destination country only when the amount rule has already fired".
How does rule version history work?
Each time a rule is saved, its previous state is kept as a version with the person who saved it and the date. You can open any version as a read-only preview and revert to it, and the revert is itself saved as a new version. Undo and redo buttons sit in the editor toolbar, so users don't need to remember Ctrl+Z and Ctrl+Y. Both work for transaction rules and customer rules.
What else changed in rules?
- Adverse media in the screening node: a rule can check whether a party's name appears in adverse media at the moment of the transaction, and the case shows which profile produced the hit.
- Counting dynamic fields: rules can count values in dynamic fields, for example comparing a merchant's foreign card transactions with the previous day.
- Customer formulas from Excel: formula rows can be imported from a template, with invalid rows listed by row number rather than stopping the import.
Why it matters. Multi-step scenarios are built without squeezing them into a single rule, and every change can be traced and undone.
AI
On more screens
The problem. AI support was limited to a few screens, and its suggestions didn't show what they were based on.
Where does AI help this month?
- Widget Builder: write what you want to see in a sentence, for example the average time it takes to close a case, and the widget is built for you.
- Network Analysis: a detection engine first scans the visible network and its transactions for shared devices, IBANs, addresses and phones. An assistant then explains what it found, points you to the unusual accounts and transactions, and suggests rules. Click a suggestion and the rule screen opens with the scenario passed to the rule assistant. The analysis is translated into your default language.
- Agentic AI at profile level: in Name Screening, agents can now decide profile by profile. Conditions can use the profile's match rate, birth year, deceased status, country and nationality and PEP class. Actions can set a profile's match status and risk level, add it to the safe list or start an adverse media scan for it.
- AI Suggest with numbers: the suggestion now shows how many similar cases your organization decided and how many were closed as true or false positives. With fewer than ten such cases, it falls back to a wider average and says so. The suggestion opens from the row in the case list and is translated into your default language.
Every AI screen carries a reminder that AI can make mistakes and that important compliance decisions should be verified, and every AI output has a feedback button.
Why it matters. AI sits next to the decision, shows what it rests on, and the decision stays with the team.
Everyday work
- Ongoing monitoring in two tabs: Name Screening and Adverse Media records are now tracked on separate tabs.
- Faster assignment: user and team lists are separate and searchable, and Assign to Me takes a case in one click.
- Dashboards that refresh themselves: choose an interval from one minute to one hour, and only the widgets refresh. Live Transactions can refresh every few seconds. A tab in the background doesn't refresh until you come back.
- Approval history: approved and rejected items stay in a new Approve Management History report with the maker, the approver, the decision and the old and new values, with filters and an Excel export.
- Sign in only with SSO: organizations can require single sign-on per user or for the whole organization.
- Customers sharing an IP: when an IP-based rule fires, the rule reasoning lists the other customers using the same IP, and the list appears in the transaction PDF too.
- Bulk crypto screening: crypto addresses can be uploaded in bulk through Excel and the API.
- KYC forms: fields can allow several options, applications can be approved or rejected in bulk from the list, and rejected applications are gathered in one place.
- Customers created with screening: tags can be set and automatic screening switched on when a customer is created, in the interface, in batch uploads and through the API.
If you already use complead, the details of each update are in the Help Center inside the app. For other months, see the Product Updates category.
