SymphonyAI Financial Services: connected financial-crime detection and investigation

SymphonyAI Financial Services supplies software for detecting financial crime and managing the resulting investigation work. Its current portfolio brings together customer due diligence, transaction monitoring, screening, payment-fraud controls and case handling. For a brokerage or wider financial group, the operational question is how customer information, transaction activity and investigator decisions move through those controls without losing the evidence behind a risk assessment.[1]

The financial-markets offering addresses capital-markets AML and monitoring of higher-risk clients, trading activity and internal actors. This makes the business relevant to institutional dealing and brokerage operations with complex customer and transaction relationships. Product selection should follow the particular activity being monitored: customer onboarding, payment screening and trade-related surveillance have different data inputs and decision owners.[2]

SRI Investigation assembles evidence for investigators and coordinates specialist agents through a managing agent. The workspace includes a case assistant, relationship and funds-flow exploration, and mandatory-action lists before closure. Teams can review the evidence, challenge findings and manage work through roles and queues. These functions are particularly relevant where an institution needs to explain why an alert was escalated, resolved or included in a disclosure.[8]

Symphony Risk Intelligence connects risk information around a shared representation of customers, accounts, transactions and relationships. It is designed to work with existing detection and case-management systems, allowing adoption by risk domain or workflow. An integration therefore needs to define the authoritative source for each entity, how identifiers are reconciled and where investigation results return. The published modular approach supports staged adoption instead of requiring every control to change simultaneously.[3]

Transaction Screening supports SWIFT MT/MX, ISO 20022 and Fedwire message formats. It separates scoring from the policy that decides whether to suppress, escalate or review a result, and supports batch and real-time list updates. Those are screening and workflow capabilities; the named formats do not themselves establish a payment-execution service or membership of a payment network.[6]

Payment Fraud describes integration with ThreatMark behavioural signals and with payment workflows that can hold, release or block activity. The implementation must connect a detected event to the institution’s authorised payment decision. Customer and payment-history signals can be considered together, but the exact upstream signals, decision latency and failure handling depend on the selected integration.[7]

Pricing is established through a service order. The publicly linked subscription agreement bases fees on the services purchased, rather than actual usage, and makes those commitments non-cancellable and non-reducible during the specified term. Its default is annual advance invoicing with payment due within 30 days, unless otherwise agreed. A proposal should identify the subscribed applications, environments, service term and implementation work, so the contractual commitment can be compared with the intended deployment.[11]

The platform’s public evaluation route offers a consultation and proposal rather than a numerical tariff. A firm comparing options needs a scoped commercial response: the applications selected, the intended data connections and the extent of migration all affect what is being purchased.[3]

KYC/CDD can use preferred identity-verification, document and questionnaire providers, with third-party information supporting beneficial-ownership analysis. Implementation includes configuring risk scores, review triggers and approval workflows. Event-driven updates can prompt a fresh assessment when circumstances change; scheduled reviews required by the firm’s policies or obligations still need to be represented. The quality of the initial customer and ownership data directly affects the usefulness of subsequent monitoring.[4]

SRI Agents provides policy checkpoints, restricted access and pause or stop controls. SymphonyAI describes evaluating agents against representative and adversarial cases and monitoring behaviour after deployment. A production design should specify what an agent may read, which actions it can take and which findings require escalation. Testing those boundaries matters as much as demonstrating that a generated case narrative looks plausible.[9]

Deployment must be matched to the product. The financial-services lifecycle policy specifically covers listed NetReveal and Sensa software in a customer’s cloud or datacentre, and excludes SymphonyAI-hosted SaaS. Those arrangements should not be treated as an identical deployment menu for every current application.[12]

The corporate security page presents assurance references including ISO 27001 and SOC 2. A finance implementation should obtain the applicable report or certificate scope, current assessment period and hosting responsibilities for the actual service. That evidence helps connect the supplier’s corporate security statements with the environment handling customer identities, transaction histories and investigation records.[15]

The financial-services lifecycle policy distinguishes standard support, limited extended support and final end of life. After standard support ends, the covered release receives no new code or security fixes; any extended support requires a separate agreement and may be unavailable. Installation assistance and alternative-product migration can involve additional charges. Lifecycle notices use the customer portal, account manager or email. The relevant release and support agreement determine the operational position.[12]

New enquiries use the corporate form with Financial Services selected. Existing customers should establish their incident, escalation and account-management routes through the engagement. Public product enquiries and legal contacts serve different purposes, so the production support schedule should identify the route used for an unavailable screening service or a failed data connection.[13]

The provider supplies financial-crime technology to institutions. Its capital-markets applications support monitoring and investigations; they do not establish the brokerage permissions of a customer using them. The institution must connect system outputs to its own decision authority, investigation standards and regulatory responsibilities. Product automation is an operating capability whose allowed actions need to be defined by that institution.[2]

SRI Investigation supports disclosure preparation, including region-specific formats and generated narratives, with human review and approval. It also allows governed handling of lower-risk cases. Firms therefore need separate policies for automatic case treatment and for approval of a disclosure: a prepared narrative, an approved case and a submitted regulatory report represent distinct stages of work.[8]

The privacy policy separates website and business-contact information from customer-submitted service data, whose processing is governed by the customer agreement. SymphonyAI can act as processor or subprocessor for that data. This distinction matters when assessing retention, access rights and international processing: a general website privacy statement does not define every term of an institution’s financial-crime deployment.[16]

The subscription agreement’s data-processing provisions require processing on documented instructions and place responsibility for lawful instructions with the subscriber. Contract review should connect those instructions with the actual customer records, third-party data and investigative material supplied to the service, including the applicable subprocessor arrangements.[11]

The financial-services business sits within SymphonyAI, a privately held enterprise-AI company headquartered in Palo Alto. Its other industry businesses include retail, industrial applications, enterprise IT and media. The group’s privacy policy identifies SymphonyAI LLC and its subsidiaries and affiliates; the customer agreement should identify the entity supplying the contracted service.[17][16]

Romesh Wadhwani founded SymphonyAI in 2017 and is its chairman. Sanjay Dhawan is chief executive, and John Edison leads Financial Services. The published leadership information places SymphonyAI within Wadhwani’s wider SAIGroup interests. Those wider investments are separate from the financial-services product portfolio described here.[18]

Symphony AyasdiAI was renamed SymphonyAI Sensa in August 2022. SymphonyAI then completed the acquisition of NetReveal from BAE Systems in October 2022, bringing the acquired financial-crime business into the same financial-services division. These earlier names remain relevant when identifying installed software, documentation and migration requirements.[19][20]

The group also completed its acquisition of 1010data in June 2023, adding decision-science, data-management and analytics capabilities serving retail and financial-services customers. That transaction extends the group’s analytics history; it does not imply that every 1010data product forms part of a current SRI subscription.[21]

Operator

SymphonyAI LLC and affiliates identified in the service order

Service scope

Institutional financial-crime detection, customer due diligence, screening, payment-fraud controls and investigation workflow.

Customer types

Banks; Financial institutions; Capital-markets firms; Wealth managers

Delivery

SaaS; Hosted; Hybrid; On premises

Pricing publication

Quote only

Headquarters

Palo Alto, California, United States

Platforms

Symphony Risk Intelligence; KYC/CDD; Transaction Monitoring; Screening; Payment Fraud; SRI Investigation; SRI Agents; Sensa Investigation Hub; SensaAI for AML; SensaAI for Sanctions

Integrations

Existing detection and case-management systems; SWIFT MT/MX and ISO 20022 transaction-message formats; Fedwire message screening; ThreatMark behavioural signals; Third-party identity and beneficial-ownership data

Pricing scope

Service-order pricing; standard subscription terms base fees on purchased services, with annual advance billing unless otherwise agreed.

Implementation

Connect customer and transaction data, validate risk policies and model behaviour, then stage the selected modules and agent permissions.

Support

Financial-services support follows the customer agreement; legacy customer-hosted products also have a published lifecycle policy.

Permissions and regulatory scope

Financial-crime software and workflow support for institutions. Product capabilities do not confer brokerage permissions or transfer the institution’s regulatory accountability.

Provider metadata checked 21 Oct 2026Source 21Source 17Source 9Source 19Source 12Source 7Source 8Source 4Source 10Source 1Source 2Source 5Source 11Source 20Source 3Source 16Source 6

SymphonyAI introduced Symphony Risk Intelligence on 24 September 2026. The launch describes a system that connects regulatory developments, threats, policy and customer behaviour, with configurable levels of agent autonomy ranging from assisted to fully autonomous workflows. The appropriate level depends on the institution’s governance choices. This is a material distinction when comparing the current platform with older Sensa or NetReveal descriptions that use different product names and operating models.[10]

The current Transaction Monitoring page presents its product as the next generation of NetReveal and describes phased modernisation alongside existing systems. It also supports customer models through model-management capabilities. A migration should therefore establish which existing scenarios, investigation records and validated models are being retained, replaced or run in parallel, with acceptance evidence for the chosen transition.[5]

The agents architecture combines deterministic rules and lookups with approved language models for tasks requiring them. It supports model choice and tracks model use, while specialist outputs can return to existing case systems. This gives an institution concrete design decisions about evidence, model access and workflow ownership; it does not remove the need to evaluate each automated task in its operating context.[9]