What we shipped in Q1: pKYB, AI, and more

Q1 was a big quarter for our agentic AI Assistant, along with a host of other product launches. We made meaningful progress strengthening our perpetual KYB solution and made exciting enhancements to our AI suite. Here are some highlights of what we shipped over the last three months.

Better business onboarding, every quarter

Staying compliant shouldn’t stop at onboarding. Two new capabilities make KYB maintenance continuous and hands-off.

Periodic reviews let you schedule re-verification of business and ownership data on a cadence tied to CRA risk tier. From assigning risk at onboarding to automating re-verification, recalculating the overall CRA, and escalating to EDD when risk increases — the entire review cycle runs without manual intervention. For best results, this should be used in conjunction with Alloy's event-driven, perpetual KYB verification.

Registry monitoring takes it a step further. Enroll businesses and Alloy automatically re-verifies identity when material changes occur, dynamically recalculates group risk, and routes cases to investigations when an agent needs to step in.

Together, they mean your KYB program doesn't just pass at onboarding — it stays accurate over time.

Deeper context for faster reviews

Fraud Signal indicators just got more powerful. 

Fraud Signal, our predictive fraud machine learning model, dynamically computes a single risk score and provides accompanying qualitative indicators that give your team a real-time representation of risk at any given time. 

Indicators are now generated at the individual evaluation level, so instead of a single static set of reasons, you see exactly what drove each score without digging through raw data. From your alerts and investigations view, you can explore the top drivers behind any evaluation, track how an entity's risk has shifted over time, and click into any indicator for a plain-language explanation.

The result: faster decisions, greater consistency, and a clearer picture of risk without relying on rules.

AI built for the way risk teams work

Manual review is one of the biggest operational bottlenecks in compliance and fraud programs. Alloy’s AI Assistant helps analysts move from review to resolution faster, with built-in explainability, so teams can trust the output and proceed with confidence. Last quarter, we brought AI Analysis to general availability and deepened the AI Assistant's capabilities across two core use cases: Watchlist Screening and KYB Review.

AI Analysis

Not every bottleneck in compliance and fraud programs comes from a complex investigation – sometimes analysts just need a clear answer to: why is this application pending, and what should happen next? AI Analysis delivers exactly that: a contextual summary of why an evaluation, application, or investigation is in its current state, generated directly from Alloy's decisioning engine and logic rather than surface-level UI text. It works out of the box based on its native connection to your rules, policies, and data sources, and its outputs can be further customized to make recommendations aligned with your organization's specific standard operating procedure (SOP).

Watchlist Screening

Resolving watchlist hits manually means hours of opening links, cross-referencing sources, and drafting notes before a single decision gets made. The AI Assistant changes that by reviewing every watchlist hit in real time – comparing applicant data against matched entities across sanctions, PEP, adverse media, and warnings sources, then surfacing structured reasoning that shows exactly what matched, what differed, and why. 

Analysts can accept a recommendation in one click, with decision rationale auto-populated and findings surfaced natively within the Alloy watchlist panel. Accepted false positives are automatically suppressed in future screenings, and every interaction is logged in a full audit trail for regulatory review.

KYB Review

KYB onboarding requires contextual reasoning across fragmented data sources that can’t be straight-through processed. The AI Assistant takes on this work end-to-end: it starts with authoritative sources for business identity verification, layers in autonomous open web and social media research with dynamic source selection, and extracts and cross-references uploaded documents like articles of incorporation, FEIN filings, bank statements, and more. It validates UBO associations and synthesizes all findings against your organization's specific SOP and decisioning context. 

The result is a recommendation that's ready for one-click approval or, for teams that want full automation, can be configured to trigger a decision without human review.

Side-by-side Document Review, now available in the SDK

We've made reviewing and managing documents faster and more efficient for teams using Alloy's SDK. If you're using one of our supported plugins, such as Veriff, Persona, Sumsub, Onfido, and more, you can now launch Entity Documents directly from the document evaluation page via a banner at the top of the page. 

When you open the documents modal from an evaluation, any supplied data will appear alongside the document, so you can easily compare document details with supplied information side-by-side. 

We've also added the ability to upload documents directly within the document window – no extra steps required. Once uploaded, documents automatically appear at the front of the document navigation for quick access. This upload capability is available for all Applications and Investigations users.

Precision controls for faster and easier Custom List management

Custom Lists let clients run self-service lookups against any data list they bring into Alloy, and this quarter, we shipped a set of enhancements that make them easier to use. 

We added support for expirations, so records are automatically removed from decisioning upon expiration, reducing stale data and eliminating the need for manual cleanup. List Row IDs introduce a unique identifier for each row so individual records can be precisely targeted without affecting matching logic. And users can now update or remove specific rows without re-uploading the full dataset, with each change creating a new minor version for traceability. Together, these updates make Custom Lists easier to maintain as they grow and change over time.

Stay tuned for more exciting updates in Q2!

We are just getting started this year. From AI-powered review to more efficiency for teams, everything we build is to help organizations move faster with more confidence. Stay tuned for more!