Pursuing such an automated decision-making nirvana can be a time-consuming trek down risky trail. Heavy IT involvement in translating business logic to decision logic is just one of the many hurdles along this trail. Companies need to painstakingly integrate and work with multiple point solutions to facilitate data flow, data analysis, prepare questionnaires for decision logic and test the decision logic.
“We saw large organizations struggle with rigid implementations that barely survived the test of time. Business analysts refused tools that looked too much like development tools. We consulted with companies that deployed decisions that did not perform as well as they hoped. Those were the problems we made our mission to solve,” says Carole-Ann Berlioz, co-founder and chief product officer at California-based Sparkling Logic, a company that flattens the transition to high speed, automated decision-making into a smooth ride, with its comprehensive SMARTS™ decision management technology.
A Decision Management Technology that is Powerfully Simple
Sparkling Logic SMARTS™ is a decision management platform that unifies the functions of authoring, testing, deploying, and improving decisions within a single tool. Rather than focusing purely on the decision logic or business rules ‘code,’ SMARTS™ focuses on ensuring that the business outcomes of the decision are measured. With data being at the core of its approach, the platform helps business analysts easily measure the business quality of the decisions as they are managing them, without requiring the help of IT teams or developers. It provides a foundation for understanding the decision logic and its effects on transactions, allowing business performance to be measured early and continuously.
What makes the solution powerful is its ability to test the decision making end-to-end. Not only can transactions be scored, but the business performance of the strategy for the proposed cut-offs can be measured for continuous improvement. Furthermore, real-time business performance is measured and reported on in the decision analytics module, without any coding involved.
At its core, SMARTS™ is the sum of four capabilities. It houses the central decision management platform which leverages data samples to intuitively author decision logics, test them, and measure the impact analysis. It is powered by a patented technology called RedPen™ that allows rule authoring without the need to learn a rule syntax. The platform features all the capabilities needed for enterprise-level lifecycle and deployment management. Organizations can share and distribute decision logic through reusable libraries, and have logic segmented based on product, client, region, or any other dimension. A low-code/no-code environment to design decision making workflows is another powerful capability within the platform. It allows users to augment decision making with an intelligent front-end and allows untrained business users to configure decision logic while enforcing governance in a no-code manner. Also, a built-in machine learning (ML) environment allows predictive models to be run as part of any operational decisions.
A global online payment platform used BluePen for fraud detection. Since the deployment of the model, the detection time of a fraudulent transaction has been reduced from two weeks to less than a day
SMARTS™ solves the challenge of model tracking, evaluation, and versioning while in production. For business applications where time is of the essence, or when models are short-lived, business experts can use the SMARTS BluePen™ learning technology to quickly create a model, potentially leveraging existing models, and put it into a production decision automation application. And lastly the platform comes with a real-time decision analytics environment to define dashboards, measure the quality of the business performance in deployed automated decisions, and to seamlessly associate metrics to business decisions. Real-time decision analytics charts display the decision performance measurements and trigger notifications and alerts when certain thresholds are crossed, or certain patterns are detected. Notifications can be pushed by email or as service tickets.
Cohesively, the prowess of these four aspects allows business analysts to intuitively author decision logic that evolves with business activity, internal policies, and industry regulations. Companies get to effectively test and measure business performance via dashboards, and decisively promote decision strategies to production or even revert to a past decision making configuration if needed. Typical business users can benefit from the agility and flexibility of decision automation. They can configure and refine their decision logic, test, and simulate decision services, experiment, and choose decision strategies, and finally publish and manage logic deployments, all through web forms and point-andclick operations. All this can be done without prior business rules skills and the need for IT or even business analysts.
SMARTS™ has been designed with ease of deployment in mind. It comes with all the required performance, security, integration, and scalability capabilities to fit any enterprise architecture and governance model. “Whether our customers have chosen to host themselves or to use our SaaS solution, we provide the most modern tooling for them and make sure we fit natively into their ecosystem. Performance and maintainability are as pleasing for the IT admin in charge of the system’s administration and operations as they are for the business analysts in charge of the decision logic,” says Carole-Ann.
Fintech Use Cases
Companies around the globe have chosen SMARTS™ as their decision-making technology of choice because out its rich, outof- box capabilities for a variety of use cases. Take assessing credit risk for example. Credit origination logic must check against ID fraud, scan for risk indicators, retrieve and process credit data, score the level of risk, and then establish the parameters of the credit offering such as credit limit, interest rate, etc. With SMARTS™, each step pertaining to these conditions can be implemented as business rules and predictive models. SMARTS™’s ModelOps capabilities offer a range of options for integrating these models. “Our customers in credit originations enjoy in particular the built-in champion / challenger capabilities. Several strategies can be deployed to compete in production, without any need to involve IT in the underlying plumbing,” says Carole-Ann. Business owners are automatically alerted when the decline rate swings out of bounds, preventing losses from accumulating due to a bad candidate strategy.
Financial fraud is another aspect that demands prompt decision making. Fraud rings have allowed fraudsters to share tips on corporate weaknesses, accelerating the spread of new tactics that are abusing these vulnerabilities. To this end, SMARTS™ expands the reach of machine learning algorithms to shorten fraud detection. As soon as transaction data is available, fraud experts can launch ML jobs that identify the new patterns that must be stopped. While data scientists focus on established patterns of fraud and refine models for accuracy, fraud experts focus on speed. “A global online payment platform uses SMARTS BluePen for fraud detection. Since the deployment of the model, the detection time of a fraudulent transaction has been reduced from two weeks to less than a day,” states Carole-Ann.
“Across the board, for credit origination, insurance rating engines, as well as any pricing engines, there is the need to calculate rates, premiums, or fees. For that purpose, SMARTS™ provides a lookup model engine that is dedicated to the fast retrieval of rates,” mentions Carole-Ann. The lookup models on the platform have significantly simplified the implementation of calculation engines. The core capability is that rates are retrieved from a rate table, rather than individual business rules. Indexing features in the platform facilitate faster processing even when there are tens of thousands of lines of rates or more. For actuaries, the key takeaway is that these rates can be imported from an Excel spreadsheet, making their lives easier. As rates change over time, a new spreadsheet can be imported, replacing the old one, or taking effect in a gradual way, once again dictated by decision logic. For instance, an American rating agency has integrated SMARTS™ into its origination platform for its corporate clients in order to manage their credit risks. With SMARTS™, the agency now carries out between 200 and 250 million transactions per year.
"Whether our customers have chosen to host themselves or to use our SaaS solution, we provide the most modern tooling for them and make sure we fit natively into their ecosystem "
Sparkling Logic accelerates how companies leverage internal and external data and models to automate and improve the quality of enterprise-level decisions. “We have seen first-hand how the pandemic has thrown a curve ball into the financial services industry. The way we used to do business two years ago is mostly obsolete and financial services companies must adapt. For some, it means incorporating new data more rapidly and more frequently. For others, it means relaxing their requirements. In all cases, it means changing their decision logic,” adds Carole-Ann. To this end, SMARTS™ provides companies with agile decision-making capabilities. They can be confident that the changes they are introducing are good for the business, and that they have the ability to measure and further improve the quality of decision making even after deploying the decision logic.
Humble Beginnings to Promising Roadmap
SMARTS™ was designed by a team of experts who spent decades in AI and wanted to deliver a powerful yet simple product so that a business analyst could start with data and build decision logic with built-in predictive data analytics and execution decision analytics. Carole- Ann, along with co-founder Carlos Serrano-Morales, are experts in product development and management, and witnessed first-hand the struggles that companies faced owing to the lack of analyst-first decision technology. Owing to the gamut of innovations and value proposition that Sparkling Logic brings to the table, many investors were quick to recognize and endorse its vision and mission. Some of the notable names in the company’s investor and mentor panel include Larry Rosenberger, former CEO at FICO—who is known for the FICO risk score used across all credit bureaus and revolutionizing credit risk—Patrick Perez, co-founder and former CEO of Neuron Data, and Charles Forgy, the inventor of the RETE algorithm for implementing rule-based systems.
With such a sound leadership and investor team, alongside novel product offering and service model, Sparkling Logic has carved its niche in the decision technology landscape. Looking ahead, Sparkling Logic continues to do what it does best: redefining the way enterprises approach decision making.
“The direction of the company remains consistent with its original vision: to provide best-in-class decision management technology. Our plans are to keep enriching the out-of-the-box capabilities for the benefit of business analysts, while keeping the tooling current in the cloud ecosystem for the benefit of IT,” concludes Carole-Ann.