CleverChain and Experian entered a strategic partnership centred on AI-powered global due diligence |

CleverChain and Experian entered a strategic partnership
centred on AI-powered global due diligence

CleverChain and Experian entered a strategic partnership centred on AI-powered global due diligence |

Learn More

| Learn More

Learn More

VERA

AI-powered Intelligence hub.

VERA

Autonomous AI Due Diligence Agent

Autonomous AI Due Diligence Agent

End-to-end Customer and Enhanced Due Diligence on legal and natural persons, in full autonomy, in minutes per entity, calibrated to each user's own policies, procedures and risk framework.

CleverScreen
CleverScreen

What it does.

From a minimal input (an entity name, or a registration number and jurisdiction), VERA validates registry information, researches the nature of business, reconstructs shareholder history and the ownership and control structure beyond registry filings, identifies significant events and company directors, performs multi-dimensional contextual screening, conducts a risk assessment against the user's risk engine, and produces a time-stamped, fully cited report suitable for the CDD or EDD file. Every investigative step is logged and exportable for regulatory review.


How it works.

VERA sits on a multi-layer intelligence engine with three configurable inputs.

First, the user's policies, procedures, use cases, regulatory mapping and internal data, pre-wired into the workflow.

Second, selected data sources, orchestrated by scenario across registries, commercial datasets, leak databases and open sources, with no dependency on any single provider.

Third, an LLM ensemble, with the best-fit model selected per task on current performance rather than a single fixed model.

The engine is policy-aware, scenario-tuned and evidence-generating. It applies agentic, hypothesis-driven reasoning to gather and connect evidence, then resolves it into a defensible narrative rather than a score.

Screening within VERA is contextual, meaning that sanctions, PEP, enforcement, adverse media and reputational signals are evaluated against geography, sector, role, chronology and network, with multi-attribute entity resolution across languages, scripts and transliterations, rather than name-against-list matching.

VERA is offered in two deployment models, both running on the same engine: full-stack, where CleverChain supplies data and intelligence; and intelligence-only, where the client supplies its own data sources and CleverChain supplies the intelligence layer.

Point-in-time and continuous. VERA operates both at a point in time and continuously. The same engine performs event-driven and time-based monitoring, re-assessing entities as ownership, control, activity, network or reputation change.

VERA combines agentic AI with a configurable orchestration architecture.

Agentic, hypothesis-driven reasoning.

Rather than executing a fixed rule set, VERA asks investigative questions (what evidence would confirm this risk, what would refute it), retrieves targeted data to test each hypothesis, updates its position, and converges on a reasoned conclusion. This is what allows a defensible narrative output rather than a score.

Agentic, hypothesis-driven reasoning.
Agentic, hypothesis-driven reasoning.
LLM ensemble.

LLM ensemble.

VERA is model-agnostic by design. It draws on multiple leading large language models and selects the best-fit model per task on current performance. This is the opposite of a standalone large language model tool, which routes every question through one fixed model. With VERA, no single model is a point of failure or a ceiling on capability, and models are used only where they outperform alternative methods rather than as the default answer to every step.

Source orchestration.

The engine integrates directly with registries, commercial datasets, leak databases and open sources, selecting and configuring sources by scenario. Direct access to structured, authoritative data means VERA does not rely on model-generated content where primary-source evidence is required.

Source orchestration.
Entity resolution.

Entity resolution.

Multi-attribute resolution combined with exact and fuzzy matching, cross-language transliteration and alias consolidation, linking the same entity across sources, jurisdictions and scripts.

Contextual, scenario-based discounting.

Each signal is assessed against context and network, suppressing look-alikes and surfacing genuine risk that list logic alone misses.

Contextual, scenario-based discounting.
Network and ownership reasoning.

Network and ownership reasoning.

Traversal of the full ownership chain, including indirect and circular structures, and mapping of directorship, shareholding and control networks across jurisdictions.

Policy-wiring and audit layer.

The user's policies, data blocks and regulatory mapping are pre-wired into the workflow; every step is logged and timestamped; and each report concludes with a control checklist mapping requirements to the evidence collected, with completion status per item.

Policy-wiring and audit layer.

VERA addresses four structural problems in KYC and KYB:

The verification and KYB gap.

The industry has invested heavily in business verification, confirming that a legal entity exists and that its registry details match official records. It has invested far less in business understanding: what a customer actually does, who controls it, and how its risk profile evolves. Verification is a snapshot of what a company is on paper at its last filing. KYB, as CDD and EDD for legal persons have always required, is the obligation to establish and maintain an evidence-based understanding of purpose, beneficial ownership and control, activity-based risk, and change over time. Most programmes stop at the first and treat it as the whole.

The verification and KYB gap.

The industry has invested heavily in business verification, confirming that a legal entity exists and that its registry details match official records. It has invested far less in business understanding: what a customer actually does, who controls it, and how its risk profile evolves. Verification is a snapshot of what a company is on paper at its last filing. KYB, as CDD and EDD for legal persons have always required, is the obligation to establish and maintain an evidence-based understanding of purpose, beneficial ownership and control, activity-based risk, and change over time. Most programmes stop at the first and treat it as the whole.

The limits of static watchlist screening.

Name-against-list matching is necessary, but it is not an investigative outcome, and two of its weaknesses are structural. First, a watchlist is a restricted universe compiled by design: the provider's analysts decide which entities merit a high-risk profile, so the list is only as complete as that editorial process allows. Risk that no analyst has profiled is invisible to every institution that relies on the list. That is a false negative by construction, not a data-quality lapse. Second, the profiles themselves are static. A high-risk profile built on, for example, a name and a past offence is a problem for the user, who cannot disambiguate it or interrogate its basis, and it is equally a problem for the provider, who is not positioned to keep it current. The profile ages and becomes stale. Watchlist screening also cannot read context: geography, sector, role, chronology, or network.

The limits of static watchlist screening.

Name-against-list matching is necessary, but it is not an investigative outcome, and two of its weaknesses are structural. First, a watchlist is a restricted universe compiled by design: the provider's analysts decide which entities merit a high-risk profile, so the list is only as complete as that editorial process allows. Risk that no analyst has profiled is invisible to every institution that relies on the list. That is a false negative by construction, not a data-quality lapse. Second, the profiles themselves are static. A high-risk profile built on, for example, a name and a past offence is a problem for the user, who cannot disambiguate it or interrogate its basis, and it is equally a problem for the provider, who is not positioned to keep it current. The profile ages and becomes stale. Watchlist screening also cannot read context: geography, sector, role, chronology, or network.

Ongoing monitoring.

Annual or triennial review cycles treat a live business relationship as a series of snapshots, leaving risk that emerges between reviews undetected until the next scheduled date. VERA's answer is continuous oversight on the same intelligence engine. For legal entities, it monitors e.g. registry and corporate status, nature of business, ownership and control, sanctions and enforcement, litigation, adverse media and reputational risks. For natural persons, it monitors e.g. role and network changes, multilingual adverse media, sanctions and watchlists, PEP status and close associates. Most importantly, the materiality of changes is interpreted in context rather than treated as a binary change, and workflows are triggered accordingly: a silent, timestamped update where a change is not material; an internal escalation backed by evidence where it is; and a client outreach workflow where action is required. Monitoring is both event-triggered and time-based, and configurable by segment, jurisdiction and product.

Ongoing monitoring.

Annual or triennial review cycles treat a live business relationship as a series of snapshots, leaving risk that emerges between reviews undetected until the next scheduled date. VERA's answer is continuous oversight on the same intelligence engine. For legal entities, it monitors e.g. registry and corporate status, nature of business, ownership and control, sanctions and enforcement, litigation, adverse media and reputational risks. For natural persons, it monitors e.g. role and network changes, multilingual adverse media, sanctions and watchlists, PEP status and close associates. Most importantly, the materiality of changes is interpreted in context rather than treated as a binary change, and workflows are triggered accordingly: a silent, timestamped update where a change is not material; an internal escalation backed by evidence where it is; and a client outreach workflow where action is required. Monitoring is both event-triggered and time-based, and configurable by segment, jurisdiction and product.

Structural dependency on a single provider.

Most offerings tie institutions to one data or LLM provider. That dependency creates exposure on at least four fronts: contractual terms, performance, coverage, and the ability to customise. VERA removes it because data sources and models are both orchestrated and selected on a scenario basis. For data, the selection is driven by e.g. the type of assessment and the jurisdiction, while for models, by e.g. the activity to be performed and each model's current measured performance. The result is scenario-based due diligence rather than a one-size-fits-all pipeline, and an architecture that is not held hostage to any single supplier's roadmap, pricing or limitations. Last but not least, this also means that new data sources can be swiftly added at any time.

Structural dependency on a single provider.

Most offerings tie institutions to one data or LLM provider. That dependency creates exposure on at least four fronts: contractual terms, performance, coverage, and the ability to customise. VERA removes it because data sources and models are both orchestrated and selected on a scenario basis. For data, the selection is driven by e.g. the type of assessment and the jurisdiction, while for models, by e.g. the activity to be performed and each model's current measured performance. The result is scenario-based due diligence rather than a one-size-fits-all pipeline, and an architecture that is not held hostage to any single supplier's roadmap, pricing or limitations. Last but not least, this also means that new data sources can be swiftly added at any time.

Experience Smarter Compliance.

Experience Smarter Compliance.

Request your demo

What sets our innovation apart

1. Our multi-layer intelligence engine

Most offerings are built around one of two things: a data asset, or a model. The latter group includes the standalone large language model providers, whose tools answer a due diligence question by generating a response from a single model. VERA's engine is different in kind, as it operates as an orchestration layer that is agnostic to both data source and models. Data sources are selected and configured by scenario, while large language models are drawn from an ensemble on a best-fit-per-task basis, and any model in that ensemble is used only for the parts of an investigation where it genuinely outperforms, with primary-source evidence retrieved directly rather than generated. A standalone model has no access to gated registry or commercial data, no orchestration across sources, no policy-wiring, and no defence against the limitations or roadmap of the single provider behind it. This removes single-provider dependency and lock-in, and it means the engine improves as the whole data and model landscape improves, without re-platforming.

1. Our multi-layer intelligence engine

Most offerings are built around one of two things: a data asset, or a model. The latter group includes the standalone large language model providers, whose tools answer a due diligence question by generating a response from a single model. VERA's engine is different in kind, as it operates as an orchestration layer that is agnostic to both data source and models. Data sources are selected and configured by scenario, while large language models are drawn from an ensemble on a best-fit-per-task basis, and any model in that ensemble is used only for the parts of an investigation where it genuinely outperforms, with primary-source evidence retrieved directly rather than generated. A standalone model has no access to gated registry or commercial data, no orchestration across sources, no policy-wiring, and no defence against the limitations or roadmap of the single provider behind it. This removes single-provider dependency and lock-in, and it means the engine improves as the whole data and model landscape improves, without re-platforming.

2. Understanding vs verification.

VERA is designed to deliver the four obligations that separate genuine KYB from identity confirmation: purpose and intended nature of the relationship, beneficial ownership and control beyond the registry layer, an activity-based customer risk assessment, and monitoring calibrated to change. It tests declared activity against evidenced activity, maps control exercised through means other than majority shareholding, and surfaces group-level risk that entity-level checks in isolation cannot see.

2. Understanding vs verification.

VERA is designed to deliver the four obligations that separate genuine KYB from identity confirmation: purpose and intended nature of the relationship, beneficial ownership and control beyond the registry layer, an activity-based customer risk assessment, and monitoring calibrated to change. It tests declared activity against evidenced activity, maps control exercised through means other than majority shareholding, and surfaces group-level risk that entity-level checks in isolation cannot see.

3. Contextual screening vs list matching.

Sanctions, PEP, enforcement and adverse media signals are re-evaluated in context (geography, sector, role, chronology and network), with multi-attribute entity resolution across languages, scripts and transliterations. The output is a graded, dated, cited narrative that includes positive and exculpatory evidence, not a hit or no-hit flag. VERA also builds intelligence on entities for which no watchlist entry exists and can be configured reputational factors that are not univocal nor can be constrained within a list or table.

3. Contextual screening vs list matching.

Sanctions, PEP, enforcement and adverse media signals are re-evaluated in context (geography, sector, role, chronology and network), with multi-attribute entity resolution across languages, scripts and transliterations. The output is a graded, dated, cited narrative that includes positive and exculpatory evidence, not a hit or no-hit flag. VERA also builds intelligence on entities for which no watchlist entry exists and can be configured reputational factors that are not univocal nor can be constrained within a list or table.

4. Ingestion of data in any form.

If required, VERA can ingest internal and external documents in any format, including PDF registry extracts, client KYC files, financial statements and corporate filings, so that material becomes evidence inside the due diligence workflow rather than a separate, manual step performed alongside it. The user's own internal data and third-party inputs are interpreted within the same investigation as registry and open-source data.

4. Ingestion of data in any form.

If required, VERA can ingest internal and external documents in any format, including PDF registry extracts, client KYC files, financial statements and corporate filings, so that material becomes evidence inside the due diligence workflow rather than a separate, manual step performed alongside it. The user's own internal data and third-party inputs are interpreted within the same investigation as registry and open-source data.

5. Interactive investigation vs a static deliverable.

Once a VERA report exists, it is not the end of the process. KIRA, CleverChain's AI digital due diligence consultant, allows the user to interrogate and extend the report conversationally, whether on content already within the report or on external data introduced into the session. A finished report becomes an investigative workspace, the way a human analyst would expect to work, rather than a fixed document.

5. Interactive investigation vs a static deliverable.

Once a VERA report exists, it is not the end of the process. KIRA, CleverChain's AI digital due diligence consultant, allows the user to interrogate and extend the report conversationally, whether on content already within the report or on external data introduced into the session. A finished report becomes an investigative workspace, the way a human analyst would expect to work, rather than a fixed document.