> For the complete documentation index, see [llms.txt](https://docs.complycube.com/documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.complycube.com/documentation/product-guides/complycube-prism-ai-tm.md).
# ComplyCube Prism AI ™
The intelligence layer within ComplyCube, combining human research, specialist AI models, and compliance expertise to support KYC, KYB, AML, and fraud risk assessment.
Prism AI operates across the ComplyCube platform. It works in the background as information moves through [Workflows](/documentation/product-guides/compliance-studio/workflows.md), checks, [risk models](/documentation/product-guides/due-diligence-tools/risk-profile.md), and [Compliance Policies](/documentation/product-guides/compliance-studio/policies.md).
It brings relevant signals into context, identifies relationships and inconsistencies, and helps the platform respond to changing risk. This supports consistent decision-making during **onboarding**, **ongoing monitoring**, and **compliance review**.
Prism AI connects information across ComplyCube’s workflows, risk models, verification checks, and compliance policies.
### An intelligence layer across ComplyCube
Compliance decisions rarely depend on a single result. They draw on information collected at different points in the **customer relationship**.
Prism AI provides an intelligence layer across this information. It helps connect what ComplyCube knows about a person or business, interpret new findings in context, and identify details that require further attention.
Its analysis informs [Workflows](/documentation/product-guides/compliance-studio/workflows.md), [customer risk assessment](/documentation/product-guides/due-diligence-tools/risk-profile.md), [policy application](/documentation/product-guides/compliance-studio/policies.md), and compliance operations. Organizations do not need to build or maintain a separate system.
### Refined through enterprise use
Prism AI has been developed and refined through the real-world use of ComplyCube by organizations operating at scale, including household-name enterprises.
These environments involve high volumes, varied customer populations, multiple markets, and complex compliance requirements. They provide practical insight into how identity and risk information behaves across different operating conditions.
This experience helps shape how Prism AI handles incomplete information, differences between customer populations, and the demands of large compliance programs. It also informs where automation can improve consistency and where findings need additional context for assessment.
### Shaped by compliance expertise
The development of Prism AI is informed by Money Laundering Reporting Officers (MLROs) and other compliance practitioners with experience managing regulated programs.
Their input connects technical analysis with practical questions: whether the available information is consistent, whether the evidence is appropriate for the risk, and whether a change in circumstances requires attention.
Compliance expertise also informs how findings are presented. An assessment should help teams understand what contributed to a result, what information may be missing, and why a matter requires review.
This combination of technical development, compliance input, and operational feedback shapes Prism AI’s approach to **KYC**, **KYB**, **AML**, and **fraud risk management**.
### Specialist models and multilingual analysis
Prism AI uses specialist models for [biometric liveness detection](/documentation/product-guides/biometric-and-liveness-verification/identity-check.md), [identity document analysis](/documentation/product-guides/identity-verification/document-check.md), and [AML screening](/documentation/product-guides/watchlist-pep-and-adverse-media/aml-screening-check.md). Each capability addresses a specific analytical task while contributing to the wider customer assessment.
#### Evolving models for biometric liveness
Prism AI incorporates certified **biometric liveness** models to help assess whether a genuine person is present during verification. These models are refined as attack methods, capture conditions, and device capabilities evolve.
Certification applies to the specific liveness technology and conditions evaluated. Model development and evaluation support continued improvements as new fraud techniques emerge.
#### Specialist training for document verification
Prism AI’s [document verification](/documentation/product-guides/identity-verification/document-check.md) models are trained for specialist document analysis. This includes recognizing document structures, assessing consistency, and identifying signs of alteration or manipulation.
This training enables detailed analysis of features that can be difficult to assess consistently through visual inspection alone. It supports repeatable examination across large volumes of documents and helps surface subtle discrepancies for further assessment.
Document findings contribute to the wider identity and risk context, allowing inconsistencies to inform the next step in the customer journey.
#### Multilingual screening, matching, and resolution
Prism AI supports the interpretation and comparison of information across languages and writing systems. This is relevant when customer details, identity documents, company records, and [screening sources](/documentation/product-guides/watchlist-pep-and-adverse-media/aml-screening-check.md) represent the same person or business differently.
Its multilingual capabilities help account for transliteration, alternative spellings, name order, and language-specific variations. It considers these differences alongside available identifying information to support matching and the resolution of potential matches.
Multilingual analysis also helps bring source material and supporting evidence into context during screening and review, allowing teams to assess findings across markets and languages.
### Continuous improvement, controlled delivery
Prism AI evolves alongside emerging risks, advances in AI, and the demands of global compliance operations. Its development combines **specialist research, compliance expertise**, and experience supporting enterprise customers at scale.
Enhancements undergo **structured evaluation before release**, assessing accuracy, consistency, and reliability across the services they support. Evaluation reflects varied customer populations, languages, markets, and operating conditions, considering both risk detection and the effect on legitimate customers.
**Controlled releases and ongoing performance monitoring** support reliable operation as capabilities advance. Applicable certification requirements remain part of the evaluation process.
This approach supports continued improvement across **verification, screening, risk assessment, and compliance operations**, with disciplined oversight of model quality.
### Hybrid intelligence and data enrichment
Prism AI combines automated analysis with input from ComplyCube’s network of human researchers and specialist partners. This hybrid enrichment model brings additional information and subject-matter expertise into assessments where available data is incomplete, ambiguous, or requires deeper investigation.
Research is ongoing, with updates incorporated several times a day. Researchers and partners contribute **source verification**, **local knowledge**, and **contextual research**. Prism AI helps connect and interpret the resulting information at scale. The availability of individual updates depends on the source and research involved.
Together, these inputs support screening, matching, and resolution, particularly where language, regional differences, or complex business relationships require closer examination. Enriched findings contribute to the wider customer or business assessment, providing additional context for risk evaluation and compliance review.
### How Prism AI strengthens the platform
Prism AI helps ComplyCube consider information as part of a wider customer or business context.
A new result may reinforce information already collected, reveal an inconsistency, or change the significance of an earlier finding. Prism AI helps identify these relationships so relevant information can be assessed together.
This includes:
* Connecting information collected at different stages of the customer lifecycle
* Identifying patterns and inconsistencies across available signals
* Interpreting information across languages and writing systems
* Supporting matching and the resolution of ambiguous finding
* Recognizing when new information changes the existing risk context
* Highlighting matters that require further evidence or attention
* Providing relevant context for compliance review
[Workflows](/documentation/product-guides/compliance-studio/workflows.md), checks, [Compliance Policies](/documentation/product-guides/compliance-studio/policies.md), and [risk models](/documentation/product-guides/due-diligence-tools/risk-profile.md) remain configurable ComplyCube components. Prism AI works across them to make better use of their information.
### Consistency at scale
Large compliance programs apply policies across substantial and varied customer populations. Differences in geography, language, customer type, and product can make consistent assessment difficult, particularly when information is collected across several stages.
Prism AI supports consistent interpretation while each organization’s configured workflows, policies, and risk approach determine how the platform responds.
Relevant context can inform how a case progresses, with the organization retaining control over its requirements and operational processes. As new information becomes available, the same intelligence layer continues to support the relationship, connecting onboarding with ongoing compliance activity.
### Explainability and governance
AI-assisted compliance decisions need visibility and oversight. Prism AI supports understandable outcomes by connecting findings with the information that contributed to them.
Compliance teams can use this context to assess why a matter was highlighted and whether further action is appropriate. Relevant information remains available alongside the applicable workflow, risk, and policy context.
Organizations remain responsible for their policies, risk appetite, and regulatory obligations. They control how ComplyCube is configured and how matters requiring judgment are handled.
### Supporting the customer lifecycle
Prism AI works throughout the customer lifecycle.
During onboarding, it places incoming information within the context of the application. As the relationship continues, it helps recognize when new information or activity changes the customer’s existing risk profile. During review, it provides context to support the assessment and resolution of findings.
By operating across ComplyCube in the background, Prism AI connects information gathered through onboarding, monitoring, and compliance operations. This provides a consistent basis for understanding risk as customer circumstances change.
### Frequently asked questions
What is Prism AI, and do I need to integrate it separately?
Prism AI is the embedded intelligence layer within ComplyCube.
It works across [Workflows](/documentation/product-guides/compliance-studio/workflows.md), checks, [risk models](/documentation/product-guides/due-diligence-tools/risk-profile.md), and [Compliance Policies](/documentation/product-guides/compliance-studio/policies.md). You do not need a separate integration.
Does Prism AI replace human compliance teams?
No.
Prism AI supports compliance teams by connecting information, identifying inconsistencies, and providing context for review.
Your organization retains responsibility for its policies and decisions requiring judgment.
How often is research information updated?
Research is ongoing. Updates from ComplyCube’s network of researchers and specialist partners are incorporated several times a day.
Timing varies by source and research activity. This cadence applies to research information. Individual customer records are not necessarily researched or refreshed at the same frequency.
How does Prism AI handle different languages?
Prism AI supports **multilingual analysis** for [screening](/documentation/product-guides/watchlist-pep-and-adverse-media/aml-screening-check.md), matching, and resolving findings.
It accounts for transliteration, alternative spellings, and name order alongside other identifying information. Language coverage depends on the capability and underlying sources.
How are Prism AI models updated and evaluated?
Models are refined as new fraud techniques and operating conditions emerge.
Updates undergo evaluation before release, followed by monitoring in operation. For certified **biometric liveness** technology, certification applies to the specific technology and evaluated scope.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.complycube.com/documentation/product-guides/complycube-prism-ai-tm.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
