AI FeaturesCopy Anchor Link

Strise bundles several AI-powered features that sit on top of the Knowledge Graph and the data you already have access to. They're designed to compress the time between opening a review and having a well-reasoned compliance decision.

All AI features are opt-in. Reach out to your CSM to enable any of them for your team.

Review AI SummaryCopy Anchor Link

Review AI Summary generates a concise, plain-language write-up of an entity's risk profile at the point of review. Instead of manually reading through every screening hit, ownership change, and flag, a reviewer sees a one-paragraph summary that cites the underlying signals.

What It CoversCopy Anchor Link

• Business purpose and principal activities, pulled from the Knowledge Graph's summary fields.
• Ownership and beneficial-owner structure, including any unusual signals (circular ownership, cross-border chains, missing beneficial owners).
• PEP and sanctions screening results, with context on the matched role or list.
• Adverse media highlights, synthesized across the latest articles in the team's configured search window.
• Calculated Risk breakdown, translated into prose.

Where It AppearsCopy Anchor Link

• Inside the review, as the first card the reviewer sees.
• Optionally in the Review PDF if your team has opted in.

Version NotesCopy Anchor Link

A V2 of the AI Summary is in staged rollout with richer reasoning and citations. It is currently internal-only while the team compares output quality -- reach out to your CSM if you want early access.

V2 runs in both compliance flows: the Assessment flow, for teams on assessment mode, and the Review card.

In the Assessment flow, V2 narrows adverse media to the hits someone has marked with Include as relevant hit in the Adverse Media Screening section. Hits left unreviewed, or marked as irrelevant, are left out, so screening false positives don't reach the summary. Adverse media is presented as its own finding and never changes the Calculated Risk class.

In the Assessment flow, V2 groups each recommended action by its basis, in this order:

• Risk context: your team's saved guidance from Risk context settings, with the applied rule quoted.
• Risk settings: the Calculated Risk points or risk class threshold behind the action.
• General advice: standard KYC/AML practice, analyst comments or adverse media, with the source named.

Risk context takes priority. Where a rule applies, the summary recommends the evidence and approvals the rule names. If a rule says to weigh a factor less, the summary does not raise its weight again. When risk context is active but no rule applies, the summary says so. Findings and suggested dispositions also state their basis, using the same three groups. Risk context shapes only the recommended actions: the risk class still comes from Calculated Risk.

In the Assessment flow, V2 closes with a Suggested disposition block drawn from a fixed set of four actions -- continue relationship, escalate to senior management, offboard, report to authorities. More than one can apply at once, each tied to a driver named earlier in the summary, and the actions that do not apply are listed with the reason. These are suggestions for the reviewer and the MLRO to weigh: the summary never states that a reporting obligation has been triggered or that a filing is required.

In both flows, V2 writes the summary of findings and a list of Recommended actions. On the Review card, the recommended actions are a short list of next steps, such as documents to request or owners to verify. Both fields can be edited before submitting, and the Review PDF shows the recommended actions in their own row below the self-assessment. In the Assessment flow, the recommended actions hold the recommendation and the suggested disposition.

In the Assessment flow, the section is called Self-assessment, the same name the Review PDF uses. It sits at the top of the walk's overview, where the summary and the recommended actions can be generated before the sections are walked; there they are read-only. In the decision step the summary can be edited and regenerated once section comments and adverse-media decisions are in. The recommended actions stay on the overview. Below the summary, the optional Completed actions field records what the maker actually did, such as documents collected or owners verified. Only the maker writes it, and generating a summary never changes it. The Review PDF shows it in its own row below the self-assessment.

AI-Powered Adverse Media SearchCopy Anchor Link

AMS AI Search is an optional layer on top of Adverse Media Screening. Instead of purely keyword-based matching, the AI layer searches broader news sources (including Google News), ranks articles for relevance to AML-specific topics, and filters out articles that are only tangentially related.

Search parameters for AMS AI -- languages, date range, keywords -- are configured separately from the classic Opoint-based AMS search so teams can tune them independently. See AMS Settings for configuration.

Assisted Document EnrichmentCopy Anchor Link

Assisted document enrichment reads documents uploaded to a company or person and suggests changes to its data, such as new shareholders, role updates or a new registered address. You accept, edit or skip each suggestion beside the passage it was read from. See Assisted Document Enrichment.

Data HandlingCopy Anchor Link

All AI features run within Strise's existing data boundaries:

• No customer data is sent to third-party model providers outside Strise's negotiated contracts.
• Results are never shared across tenants.
• The same GDPR and SOC 2 controls that cover the rest of the platform apply here. See Compliance in Strise for details.