Most alerts an analyst reviews look a lot like alerts the team has already decided. The same kind of namesake, the same list, the same reason for closing. That history sits in the case records, but each new review starts from scratch, and the knowledge of why a similar case was closed lives in the heads of the people who closed it.
Onboarding had a different problem. Customer data arrived by email, in spreadsheets or through a separate form tool, and someone had to type it into the customer record and attach the documents by hand. And when a rule found something important, the result stayed inside the platform until someone noticed it, even when another system needed to act on it straight away.
We built the May release around these moments. The highlights:
- AI Suggest proposes a decision for name screening cases, based on how your organization decided similar ones.
- KYC forms are designed in complead, shared by link and turned into customer records.
- The AI assistant on the Canvas Rule Editor understands longer and more complex requests.
- A Trigger Webhook step lets a rule notify your own systems.
- Bookmarks in the left menu keep your most used pages one click away.
AI in case review
AI Suggest for name screening
The problem. Analysts spent a large part of the day on alerts that resembled cases already closed, and the reasoning behind those earlier decisions wasn't visible when the next alert arrived.
What changed. AI Suggest brings your team's past decisions into the review.
How does AI Suggest work?
When you open a name screening case, AI Suggest looks at how your organization decided earlier cases with similar characteristics, such as true positive and false positive decisions on comparable matches. From that history it proposes a match status for the current case, with a confidence score that shows how strong the pattern is.
The analyst stays in charge. You can accept the suggestion in one click or reject it and decide differently, and the decision is recorded as yours. AI Suggest is switched on per organization, so each team decides when to start using it.
What does it look like in practice?
Picture an alert on a common name with a partial match on a sanctions list, where the birth year and nationality don't line up with your customer. If your team has closed many alerts like it as false positives, AI Suggest proposes False Positive with a high confidence score. If the history is mixed, the confidence is lower, which is a signal to read the profile more carefully. The example is illustrative, but the logic is the point: the suggestion is only as strong as the pattern in your own decisions.
Why start with past decisions?
A suggestion based on your own decisions reflects your own risk appetite. Two organizations can look at the same namesake and reach different conclusions for good reasons, and a suggestion built on each team's history respects that. It is also the first step toward agents that can handle well-understood cases on their own, under rules the team sets.
Why it matters. Repetitive alerts move faster, decisions stay consistent across the team, and every decision still belongs to a person.
Onboarding
KYC forms you design
The problem. Collecting customer information outside the platform meant retyping data, chasing documents and keeping the onboarding record in several places.
What changed. KYC forms are now designed and published in complead, and submissions become customer records.
How do you build a KYC form?
You design the form in the panel, with the fields your process needs for individual or corporate customers. Field types such as text, numbers, dates, dropdowns, checkboxes and file uploads mean a form can ask for anything from a date of birth to a copy of an identity document. Once the form is ready, you publish it and share its public link with your customer.
What happens when a customer submits the form?
The submission arrives in complead with its data and documents. After a reviewer approves it, a customer record is created automatically, the answers are written into the profile and the uploaded files are saved to the customer's attachments. The analyst doesn't retype anything, and the onboarding evidence sits on the same record that will be screened and monitored later.
Why keep onboarding in the same place as screening?
When a form and a screening tool live apart, the link between what the customer declared and what the screening found has to be rebuilt by hand during an audit. With the form in complead, the declaration, the documents and the later screening results belong to one customer record. A reviewer approving a submission sees what was sent, and the approved data is the same data that ongoing monitoring will use from then on.
Why it matters. Onboarding starts where screening and monitoring already happen, with one record and a clear trail from the first form to the first review.
Rules and automation
A stronger AI assistant on the canvas
The problem. The AI assistant on the Canvas Rule Editor handled short requests well, but longer scenarios with several conditions needed to be split up and built in pieces.
What changed. The assistant now handles more complex requests and draws the full rule tree.
What kind of request can it handle?
You can describe a scenario with several conditions in one sentence, for example "raise an alert when a customer under 25 sends more than three transfers to different countries within a day and the total is above a set amount". The assistant draws the whole tree on the canvas, with each condition as its own node. You check it, adjust a threshold if needed and save. As before, the rule is fully visible and editable, the assistant only does the first draft.
Trigger Webhook
The problem. When a rule fired, other systems learned about it only if someone passed the information on, which is too slow for actions like holding a payment or contacting a customer.
What changed. A rule can end with a Trigger Webhook step that calls your systems directly.
How does Trigger Webhook work?
First you define a webhook on the Webhooks page, with its address and a secret key. On the canvas, you add a Trigger Webhook step at the end of the rule, pick the webhook and choose the fields to send, such as values from the transaction, the customer or the account. When the rule fires, complead sends a POST request with that payload, signed with the secret key so your system can check where it came from.
A typical use: when a customer crosses a risk score threshold, the webhook tells your core system, which can send an SMS, open a ticket or call the customer, without anyone copying data between screens.
Why it matters. Rule results turn into action in the systems that need them, at the moment the rule fires.
Everyday work
Bookmarks in the left menu
The left menu has a Bookmarks area. Pin the pages you use most, and they appear at the top of the menu. Open Manage Bookmarks to drag them into the order you like. Bookmarks are saved per user, so they follow you to another browser or computer.
Smaller changes you'll notice during the day
- Shared variables: a value such as an age or amount formula can be defined once as a variable and used in several rules.
- Infinite scroll: Name Screening, Ongoing Monitoring and KYB results load as you scroll, instead of through a Load More button.
- Clone rules and rulesets: a rule or ruleset can be copied in one click as the starting point for a new one.
- AI case summary in adverse media: the AI case summary used in name screening is available in adverse media cases.
- Generate AI Note in more places: transaction case closing reasons and memos, closed customer cases and account memos can be drafted with AI.
- Withdraw an approval request: a Withdraw Management tab lists the actions you sent for approval, and you can withdraw a request before it is decided.
- Readable numbers on the canvas: numeric inputs show thousand separators, and missing values are flagged before you save.
- Max Match Rate in batches: name screening batch details show the highest match rate for each row.
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.
Previous month: April 2026 updates. Next month: June 2026 updates.