For many compliance teams the day starts with a search. You have a reference number, but first you need to pick its type. The adverse media results for a customer in the news are full of pages unrelated to the case. A threshold is about to change, and it isn't clear who should approve it. And there is a newer question: if your customer deals in crypto assets, who owns the address on the other side?
The August release was built to answer these questions. The highlights:
- Crypto addresses can be screened against sanctions lists, and matches become cases.
- In adverse media screening you choose the sources, and an optional AI review removes irrelevant news.
- The rule engine gained asynchronous rules, counter reset, a day filter and a choice of which alert levels become cases.
- The Simulator was rebuilt around before and after comparisons.
- Approval rights are defined per screen and per action.
- One search box finds any record without choosing a type.
Crypto and sanctions
Crypto address screening
The problem. When crypto transfers sat outside sanctions screening, the risk stayed out of sight, and checking addresses by hand left no trail in the case.
What changed. Crypto addresses are screened in the same search box as names.
How does crypto address screening work?
You paste a crypto address into the search box, the same box you use for names, and the address is screened against sanctions lists. A clear result says so plainly. When the address belongs to a sanctioned entity, you get a match screen in the layout you already know from name screening, with the entity's details, aliases and the lists it appears on.
Two parts of that screen matter most:
- Linked To: the people and organizations the sanctioned entity is connected to.
- Other crypto addresses of this entity: the rest of the addresses that belong to the same entity.
Both are clickable, and clicking one starts a new screening. You can start from a single address and work through the connected structure step by step, without copying addresses between tabs.
What happens to a crypto match?
A match becomes a case like any other. It can be assigned to a user or a team, carry attachments and notes, and use Generate AI Note to draft the reasoning. The case shows up in the same lists and reports as name screening cases, so crypto risk is reviewed in the workflow your team already follows.
Why it matters. You can start from one address and work through the connected structure step by step, and the whole investigation stays in the case.
Adverse media screening
Source selection and an AI review
The problem. When irrelevant news gets into the results, data quality drops and analyst time is lost.
What changed. August gives the team two filters: one the system applies, and one the team controls.
How does the AI review layer work?
Adverse media screening pulls articles from many sources, and some of them mention the name without saying anything about risk. An AI review layer now checks the headline and content of each article one last time before the results reach you, and removes the ones that carry no risk. It is an organization setting, called AI Relevance Filtering, and you can switch it on or off.
How do you choose the sources?
Every scan has two source lists. Allow Sources restricts the search to the sources you pick. Block Sources keeps the sources you pick out of the results. You can search for a known source or type a domain. Leave both empty and the scan covers every source as before.
An administrator can set default source lists for the whole organization, and an officer can still override them for a single scan. That keeps the policy in one place without blocking the analyst who needs a wider search on a specific case.
What else changed in the results?
- A summary instead of a list of headlines: the summary in the right-hand panel now explains what allegations brought the person into the news, the outline of the events and why the name was flagged.
- Several languages in one search: you can filter news in up to three languages in the same scan.
- Older news on demand: Load More goes past the first set of results in each category, and new results keep the date order.
Why it matters. The analyst spends time reading risky news rather than cleaning up the list, and the source policy is set once for the whole organization.
Transaction monitoring rules
Async, Reset Counter, day filter and more
The problem. Heavy rules could hold up the API response, and cumulative rules could produce alert after alert for the same limit.
What changed. The rule engine gained several controls.
What does Async do?
Each rule in a ruleset can be marked as synchronous or asynchronous. A synchronous rule is evaluated in the real-time response, as before. An asynchronous rule runs in the background through a queue and adds its score to the transaction once it completes. The API response time is then shaped only by the rules that must answer in real time, even in a crowded ruleset.
What does Reset Counter do?
Cumulative rules add up amounts or counts over a period, for example a daily total. Without a reset, every transaction after the limit is crossed can trigger the same rule again. With Reset Counter, the counter goes back to zero once the limit is exceeded, and the next alert comes only when a new cycle fills up.
Which other rule controls arrived?
- Day filter: a rule can be limited to certain days, for example only Saturday and Sunday for a rule that measures weekend behavior, without touching its time period settings.
- Levels that become cases: you choose which alert levels are converted to cases, anywhere from Very Low to Critical. If you choose none, the previous behavior of High and Critical stays.
- Comparison with the sector average: a merchant's turnover can be compared with the average turnover of its own merchant category code, as a multiple, without adding each code to the rule by hand.
- More precise time windows: rules can look at the last full week, month or year, and a relative window can leave today out, for example the 30 days before today.
- Party-specific IP checks: rules can look at the originator's IP address and the beneficiary's IP address separately.
- AI in customer formulas: describing a rule in a sentence and letting AI build it now works for customer risk formulas too.
Why it matters. Response time is shaped by the critical rules, and meaningful alerts reach the queue.
Simulator
How do you test a rule before switching it on?
The Simulator replays historical transactions through a rule or a ruleset you choose. You pick the transactions to replay, see how many there are before you start, and run the simulation. The result shows the alert-level distribution before and after the change, and the outcome for each transaction. A "changed only" view lists just the transactions whose outcome would differ.
Live screening isn't run during a simulation, so no screening credit is spent. A short guided tour walks through the screen the first time you open it.
Permissions and approvals
Approval matrix
The problem. Approval rights were defined per user, so a user needed approval for every action or for none.
What changed. Approvals now follow screens and actions.
How does the approval matrix work?
Approval rights are now set per screen and per action. The list covers the actions a compliance team cares about, such as creating, changing or disabling rules and rulesets, alarm thresholds and settings, case status, match status and risk level changes, customer and account blacklists and whitelists, and customer operations.
For each action, a user gets one of three behaviors:
- Perform the action without approval.
- Perform the action, which then waits for approval.
- Approve other people's requests for that action.
When you click an item waiting for approval, a detail window shows exactly what changed, with the old and the new value, and approval or rejection happens in the same window.
How do teams fit in?
Users who aren't admins only see and act on approval requests from their own team. Admins keep access to every team. Team members are listed one per row, and a member's role can be changed independently of the team's role. A single person inside a team whose actions go for approval can act as the team's approver, for example.
Why it matters. The four-eyes principle follows the real structure of the team and can be traced in an audit.
Everyday work
One search box
What does the search box find?
You no longer choose a type before searching. Type a scan ID, a reference number, a customer number, a transaction ID, a customer name, a rule name or even a menu item, and the search takes you to the right record. Click the empty box and your most recent cases and customers appear as suggestions. Results respect team visibility, so a user only reaches records their team can see.
Smaller changes you'll notice during the day
- Widget data: downloadable as Excel, CSV or HTML, with a description on each widget.
- Resolution time: a Widget Builder metric for how long cases take to close, with average, minimum and maximum.
- Company search by person: every company where a person is a director, officer or shareholder in one list.
- Files tab: files from scans, cases, notes and batch uploads in one list on the customer detail page.
- Bulk updates in ongoing monitoring: key fields updated in bulk from the interface and the API.
- Transaction details: applied and triggered rules shown separately, with a Rule Details column in the lists.
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.
