FICO: decision software, fraud operations, business rules and optimisation

Falcon combines consortium-derived models with options for custom models or an institution’s own models. The product also supplies rules, investigations and customer-contact capabilities. A model score can therefore feed a broader response workflow: policy evaluates the risk, operations investigates the case and communication reaches the customer. The product description covers several payment channels; a specific implementation still depends on the feeds, models and controls selected for that channel.[2]

Blaze Advisor can incorporate models expressed through PMML and common analytical languages including Python, R and SAS. Its deployment wizard produces decision services that can run independently or be integrated with existing systems. Authoring and execution remain connected through testing, tracing and historical change records. This makes a policy update an identifiable software change, with an execution path that can be inspected when an outcome is challenged.[3]

The Xpress family separates the optimisation engine from the surrounding application. Solver executes the mathematical problem; Mosel expresses models; Workbench supports development; Insight presents scenarios and results to business users. The Compute Interface connects optimisation with other applications. Native language APIs allow model inputs such as forecasts and coefficients to enter an optimisation workflow, while dashboards expose the trade-offs in the resulting choices.[5]

There are distinct Python integration routes. The Xpress Insight developer guide describes applications written in Python through the xpressinsight module. It explicitly distinguishes that module from the python3 extension used to execute Python commands from Mosel. These are different development environments, so an existing Mosel integration and a pure Python Insight application do not share an interchangeable package setup.[10]

FICO’s December 2025 AWS announcement identifies three additional private offers: Credit Line Optimizer, Loan Pricing Optimizer and Platform Omni-Channel Engagement. The first two formulate trade-offs between lending objectives and constraints; the communications capability supports customer contact within rules and multistep workflows. Marketplace availability supplies a commercial route to those products, not evidence that every FICO product or pre-existing licence has the same cloud entitlement.[8]

The public materials establish several licensing routes rather than one price for the portfolio. Production software and cloud services follow their commercial contracts, while Xpress also has free Community and Academic options and a complimentary evaluation route. The published Xpress limits are especially relevant when moving from a small model to a production-scale application.[5]

Professional services are separately scoped in a statement of work. The published policy distinguishes time-and-materials work, invoiced monthly against actual hours, from fixed-fee work invoiced according to progress or milestones. Time estimates are non-binding. Training scope and fees depend on the course and delivery; prepaid training must be completed within one year. Travel and associated expenses are additional under the policy.[11]

Platform decisioning supports Decision Model and Notation, decision flows and several forms of business logic. The documented sequence moves from defining inputs and dependencies to authoring rules, testing outcomes, deploying and comparing alternatives. Champion/challenger execution and simulation can compare a changed policy with the incumbent. FICO also describes CI/CD integration, connecting the business-owned decision model with a controlled software release process.[4]

Xpress Executor provides a concrete example of those controls. Its documentation requires a separate component for each model. Model and execution settings can be changed only in Design, then promoted through Staging and Production. The configuration includes concurrency, an execution timeout and whether run logs are written to S3. An update therefore passes through the lifecycle again rather than editing the production component’s model in place.[9]

The professional-services policy distinguishes a FICO-led implementation with defined requirements, design and testing from client-led work or staff augmentation. New work outside the agreed scope requires a signed change order. It also retains FICO ownership of the service deliverables while granting the customer the limited use described in the agreement. Project responsibilities, acceptance work and rights in custom deliverables therefore need to be understood together.[11]

The Global Support Center is available around the clock for customers with an active maintenance contract. Its public instructions require Severity 1 and Severity 2 production incidents to be raised by telephone; other cases can use the online portal or support email. The product and applicable policy determine the response entitlement.[14]

Unless the subscription states otherwise, the 99.9% monthly uptime target covers real-time decision components in production, excluding authoring, design, provisioning and reporting. Scheduled maintenance and outside factors are excluded. Credits require production eligibility and a claim within 30 days after month-end.[12]

Software maintenance also depends on product lifecycle. Supported releases, third-party platforms and browsers matter, and some retired releases may require paid extended support. Customers must nominate trained primary and secondary contacts and provide diagnostic information. FICO Learning supplies product-specific training, including Falcon case handling and rule authoring, Decision Modeler development and CCS operations. Its foundational Platform certification is publicly offered free.[13][15]

SaaS use is limited to authorised users, agreed territory and permitted internal processing. Personal data is prohibited in non-production environments; trials are for evaluation.[12]

The Data Processing Addendum identifies FICO as processor and the client as controller. The client determines the submitted data and must have the required disclosures, consents and authority; sensitive personal data needs prior notice and FICO authorisation. At contract end, the addendum provides for return or destruction where the client cannot retrieve or delete the data itself, subject to contractual or legal retention and the normal backup-destruction process.[17]

The SaaS policy schedules irretrievable client-data deletion 30 days after expiry or termination, subject to its security agreement and DPA; customers can download data at their expense or request return. Its licence for aggregated or depersonalised data survives the agreement for product development and other stated purposes.[12]

FICO’s Responsible AI offering includes methods for bias detection, explanation, model monitoring, governance, simulation and robustness testing. These support the institution’s model-control process. Their presence does not establish that a particular credit or fraud strategy is unbiased, regulator-approved or suitable for automatic execution without the institution’s own policy and validation. The substantive boundary is the model, decision logic, data and operating authority configured for that use case.[16]

FICO is the brand of Fair Isaac Corporation. Its quarterly report filed on 29 July 2026 identifies a Delaware corporation with principal executive offices at 5 West Mendenhall, Suite 105, Bozeman, Montana 59715, and common stock listed on the New York Stock Exchange as FICO. The report separates Scores from Software and describes both SaaS and on-premises software activity. Product-level delivery still depends on the purchased offering.[18]

Siron is an important historical boundary. IMTF announced completion of its acquisition of the entire Siron Compliance and AML business on 22 November 2022, including products, employees and customer contracts. Fair Isaac Germany, formerly Tonbeller, became IMTF’s German entity. Siron and TONBELLER names encountered in older FICO material therefore belong to that transferred business and are not presented here as current FICO software offerings.[19]

Operator

Fair Isaac Corporation

Service scope

Enterprise decision software, fraud analytics, business rules and mathematical optimisation, alongside a separate credit-score business.

Customer types

Banks; Payment providers; Lenders; Enterprise risk teams; Fraud operations teams; Analytics and operations-research teams

Delivery

SaaS; On premises; API

Pricing publication

Public and quoted pricing

Headquarters

5 West Mendenhall, Suite 105, Bozeman, Montana 59715, United States

Platforms

FICO Platform; Falcon Fraud Manager; Blaze Advisor; Decision Modeler; Xpress Solver; Xpress Insight; Xpress Workbench; Xpress Mosel

Integrations

AWS Marketplace private offers; PMML; Python; R; SAS; Xpress Compute Interface; Xpress Executor; xpressinsight

Pricing scope

Commercial software and service contracts; Xpress also provides free community and academic routes with distinct limits.

Implementation

Define decision inputs and policy, validate outcomes, integrate decision services and promote controlled model changes into production.

Support

Global Support Center, regional telephone routes, online cases and product training; SaaS and licensed-software support policies differ.

Permissions and regulatory scope

Software and analytical outputs supporting institution-defined credit, fraud and operational decisions; data processing and permitted use follow the relevant agreement.

Provider metadata checked 21 Oct 2026Source 1Source 2Source 3Source 4Source 5Source 6Source 7Source 8Source 9Source 10Source 12Source 13Source 14Source 15Source 17Source 18

PostFinance provides a named example of linking fraud detection with customer contact. FICO’s 6 November 2025 announcement describes existing Falcon use for debit cards and an expanded relationship involving Customer Communication Services. Bank executives describe the need to reach customers promptly when fraud is suspected. The announcement combines an existing relationship with language about extending usage, so it establishes the reported operating context and expansion plan rather than a separately verified completion date for every added capability.[20]

The May 2025 AWS announcement also quotes Westpac New Zealand’s existing use of FICO Platform. The bank describes moving from individual account decisions toward managing customer relationships across accounts. That case illustrates why shared customer state matters in the platform design: decision services can draw on a broader customer context while the bank defines the strategy being executed. It is a customer statement published by the vendor, not a universal performance result.[7]