Catalyst Challenges

Title: Proactive cyber security platform

Context: Data security is contemporarily one of the most dreading (dreadful) challenges. The data which goes through internet browsing and phishing are certainly under radar and most important due its magnitude. At times data is sent out even unintentionally on a phishing attempt.

Description: To create a platform that provides preemptive alerts to users on the first click prior to opening the destination (webpage), Capability to blocking an ongoing attack and statistical reports.

Expected Solution: We are looking for a solution preventing phishing attacks, presenting the following features: - Proactively scan a website and throw stats upfront before opening the website- reporting module for Risk Correspondence: (Web-Statistics, tracking of information shared in GDPR context, User statistics and connecting these anomalies with existing platform to perform risk analysis - White listing & Blacklisting websites in click of a button.

Title: Fraudulent document detection

Context: The KYC team spends a lot of time to detect if a document is genuine or if any part of it has been photoshopped. The need is for a tool to identify the genuineness of the document.

Description: Tool to recognize falsified documents, alert and build a comprehensive report to the users for fraud detection. To ascertain that the document is 100% original and that it has not been tampered with using photoshop or any other such software’s.

Expected Solution: Detect the frauds of submitted document. Check if the document is fully original and not tampered with. Or that data is not borrowed from anywhere else

Title: Web application exposing all new vehicles in stock

Context: Currently some customers of our car leasing and fleet management division cancel their orders as they are in an emergency or cannot wait any more. The ALD Automotive leasing team needs a web-based application that can help customers view all new vehicles in stock thereby allowing them to make a wiser decision for quicker delivery.

Description: Create a website application for all brands of automobile manufacturers to enlist all new vehicles in stock. The purpose is for delivery teams to propose a vehicle set readily available to the customer. Car manufacturers and car dealers can update the platform, as well as web scrapping.

Expected Solution: Web application updated by automotive stakeholders (OEM, dealers) indicating vehicles readily available in stock with details, colors, accessories, delivery time, photos, seller coordinates, etc.

Title: Audit planning Tool

Context: Audit department has multiple teams across countries with different expertise and large number of missions defined, the need is for a tool to help optimize

Description: Each year Audit control department defines a set of missions allocated to the number of audit teams. Each mission is led by one Audit team, with the possibility of other teams to contribute. This method of functioning is significantly increasing (more and more contributions on missions). It turns the scheduling of missions more and more complex due to the synchronization of team’s availability. Currently it is time consuming and a real brain teaser for managers to construct an optimized resource scheduling without an appropriate tool. The tool should to be centralized and give the possibility to managers to easily exchange on their constraints / availability.

Expected Solution: A web-based application to address these: Scheduling with a defined number of man days allocated to teams. Assigning mission to auditors. Ability to add constraints to missions (ex: must start after a defined date, speaking languages, required expertise, required certifications, seniority of auditors…) The application to also propose an optimized scheduling considering constraints of teams.

Title: Data Retention Management

Context: In line with EU GDPR, is it required to define data retention periods to ensure data is not being stored / captured longer than necessary timelines.

Description: There is a need to identify data sources across emails / servers / applications for user’s personal data in real time, capture data, date of creation of the data, type of data. Then we have to assign a retention period to trigger data deletion when time is due to delete.

Expected Solution: Ability to be alerted of all such data sources for a user to decide on deletion or not, with these features - Data sources identified on dashboard; Listing/ Pop ups of triggers for retention period; Data deletion confirmation / audit trail.

Title: Consent management

Context: In line with EU GDPR, is it required to define and capture consent from data subjects to confirm they authorize their data to be stored.

Description: There should be a feature / platform to collect individual’s consent in way that the users are aware of the where all their data is stored across emails / servers / applications and with the ability of the firm to confirm reason against every identified data source. Also, with a facility to withdraw consent for either on one or more or all data stored.

Expected Solution: Solution enabling the management of data storage consent on need basis. Retrieve and store data subject’s agreement for the data to be stored and shared within the firm

Title: App to 'quick fix' windshields

Context: Car windshields get damaged very often and full replacement comes at a high cost. The need is to provide a platform to car leasing customers so that they can report the damages in real time and get an automatic appointment for repair.

Description: Customers suffer a windshield damage on their leased vehicle year on year, they wait too long to repair it and end up replacing the whole windshield. It would have been cost effective if they had made the repair as soon as the impact was identified.

Expected Solution: The 'quick fix' App should enable the driver to take pictures of the impact on the windshield which triggers a series of events (notification to the supplier network based on ranking by ALD/Société Générale and location of the driver, appointment proposal by supplier with push notification, button to accept the appointment or decline/ask to be called back by the supplier, confirmation of appointment, push notification to remind the driver a few hours before the appointment)

Title: Centralized Digital Assistant for all HR needs

Context: Modern HR work doesn’t always happen at a desk, or even in an office. We need an HR Mobile App that can keep up with your busy schedule. Need is raised based on the market practices, employee questions, digital transformations and employee experience.

Description: A single source of connection between HR (applications) and staff. The amount of time spent by staff, managers on MyHR and HRCO is massive for executing administrative tasks (with no live chat). Within HR department, there are multiple ERP/tools to be accessed by staff to request letter, claim an expense, view pay slips, trainings, evaluations, etc. And likewise, for manager to provide approvals on each of these tools. There is a time lag between getting the right information on policies, procedures, entitlements (leave balance, travel), benefits, evaluation status, list of regulatory trainings to be completed, etc. and the administrative work involved to support all staff in group. Know your team - Analytics for HR & Managers, (Employee Directory, Team Calendar).

Expected Solution: NEO-MyHR, an all-in-one platform with – A chatbot for regular queries & live chat with MyHR - instant visibility on leave entitlements, team calendar, employee directory, ESS and MSS functionalities. Employees taking a training from mobile/smart phone. Employees applying for leaves, request for letter, probation forms, employee directory, organizational chart, view staff profile on JIVE, MyHR and other connect information. NEO API’s Should connect with PS-Expense tool to help on claiming expenses, have case number and audit trails for reference and accept IJP application and interface them without HR central tool. Solution also has to be GDPR compliant.

Title: Tool to filter over/under invoicing

Context: Money laundering through the over and under invoicing of goods and services is one of the methods of fraudulently transferring money across borders. The key element of this technique is the misrepresentation of the price of the good or service to transfer additional value between the importer and exporter. By invoicing below the “fair market” price, the exporter can transfer value to the importer. Alternatively, by invoicing the good or service at a price above the fair market price, the exporter can receive value from the importer.

Description: A proactive, well-planned approach will help us curb this money laundering menace and ensure regulatory compliance. Unlike conventional AML compliance monitoring, the Invoice-based screening is impacted by lack of readily available, reliable and adequate information to identify money laundering transactions. To overcome this challenge, we need to do analytics-driven profiling of customers to monitor their trading practices and identify anomalies in their trading behavior, like a sudden surge in the pricing or traded amount, or change goods traded, or trading with unrelated counterparties.

Expected Solution: To Build a tool which can provide advanced analytical and statistical data on the Goods traded, Origin, Pricing Dynamics etc.- Tool should analyze the data, along with other customer information such as their line of business, the kind of goods they trade in (high, low, or medium risk), and the market price of such goods.- Based on the customer profile, a database can be created which lists all the traded goods. Additionally, monitoring the prices of traded goods can also facilitate the compliance process. - Reports can be prepared to keep a track of the difference or hike in the price of the products or goods, can be prepared and published on a quarterly basis.- Data pooling, particularly data from customs and the freight community.- Further, using various data from external/social media and web analytics can publish the current or forecasted prices of goods.- Caution list on suspicious trading or invoicing activity customer wise can be prepared, which can serve as a screening tool for further trade transactions. - Build a Proactive filtering based on different past Incidents (Case Studies) reported across the Banks in the Globe.

Title: Personalized reporting module for Cash management portal

Context: Sogecash Web is the Société Générale Cash Management international portal. Global Cash is the online banking module for international corporates integrated to the SG Markets platform. The need is to build a personalized reporting tool on top of these 2 solutions.

Description: We want to add the existing reporting dashboard some personalization features, identified as priority demands. Mockups and test modules will be presented to to preselected clients (beta-testers).

Expected Solution: The main expectations regarding enhancements in the Global Cash reporting module are the following: - a customization of their display in Global Cash allowing for example the ability to choose the data displayed, aggregate selected accounts as well as virtual accounts (account references within the real account) - a graphical representation of their accounts (personalized data visualization), - the possibility to export their data in a personalized way (in excel or any other format, choosing the data downloaded for example)

Title: Tool to Automate tests & UAT

Context: Heavy life cycle is involved in creation of non-regression suites for large cross-Asset trading platform. The expected tool should perform auto generation of large scale test automation programs from natural language / Voice to text inputs.

Description: Large scale cross-Asset application deployments may generate regressions causing disruptions to the front and back office trade life cycle activities. Process of automation is highly manual and lead time to development of new regression tests is not fast enough to match the development of new features.

Expected Solution: The solution should be able to launch applications, scan the user interface and perform tests like humans based on the auto generated test programs.Tool to be built based on NLP plus Voice to text for understanding test cases.AI based programming to generate the test scripts - Screen scraping tools like Automation anywhere to perform the tests.

Title: Tool to know employee needs

Context: The groups human resource department needs a tool to gather insights and analyze the needs of employees through the tickets raised.

Description: Quite often, tool evolutions are chosen without having insight of what final users want. The solution should analyze the ticket addressed to the support team to correlate frequently asked question and offer a suggestion box to the users. It permits to capture what would bring value to the users in their usage of the tool and it is shared with application owner to pilot evolution of the tools

Expected Solution: Jump ticket data analysis with suggestion box using Big Data, Machine learning, Data visualization and Sentiment Analysis

Title: Business Analytics and Insights Engine.

Context: To build a predictive analytics tool for large volume of lending and leasing data.

Description: We have large volume of golden source lending and leasing Commitments/Contracts data for both retail and non-retail customers. We would like to power our users having golden source data with Predictive and Business Analytics.

Expected Solution: Leverage on our global datasets and arrive at an insightful Business Analytics Models that would help our Finance and Risk team to make use quickly for their reporting and regulatory needs. Create Predictive Analytics Models - based on 360-degree customer, accounts data and their current contracts, exposure; and suggest Cross and Up-sell of our other business products effectively.

Title: Predictive Analytics tool to monitor Credit Risk

Context: Risk and accounting data are obtained in one big application from multiple applications with more than 100 functionalities. This application is responsible for generation of credit risk data by performing many production activities with time crunch. This application generates lot of operational and functional data that can be used in several ways to improve performance, predict data quality issues and reduce the time to generate quality data.

Description: Tracking of availability of data from application, existing reports & external interfaces, re-engineer the data, analyze the data and identify bottlenecks. Use predictive analysis of functional and technical data to bring various solutions, especially on reducing incidents.

Expected Solution: A tool to check the restitution usage, web analytics/usage reports for Safir UI and providing customizable data comparison reports.- Tracking of availability of data in different reports & external interfaces and identify bottleneck - Predictive analysis of functional and technical incidents.

Title: Breaks reporting and reconciliations automation.

Context: Publishing the group financial communication and the regulatory reporting is an elaborate process. The reconciliation reporting period is very less (7-8 days) and highly stressful for the business. The need is to automate the entire process. Manual activities on the excel takes up 2 to 3 days (~40% of time).

Description: Automatic identification of breaks during the reconciliation process in internal tool SAFIR. Breaks are the differences between accounting and risk data that needs to be corrected before financial reporting.

Expected Solution: A configurable tool that automatically extract relevant reports, arranges required data, pivots, performs provided calculations, analyzes as per stipulations, highlight breakdowns and generate intelligent reports. The tool should be flexible for users to input the required actions before data arrives and remove all the manual activities done on excel.

Title: Increase the accuracy of Credit Risk portfolio management.

Context: Regulators force all banks to increase their maturity in term of Credit Risk Management. RISQ direction through the CORISQ committee is performing the Credit Risk portfolio management with BL(s).Some sensitive portfolios need a strong follow up and the accuracy of its perimeter is very important to handle "Concentration Risk".

Description: The business is reviewing manually for example the gap between a Corporate portfolio and a following portfolio by comparison at the third party level in terms of number of PM, amount by currency, the objective being to evaluate what is outside of the following portfolio and evaluate if it should be inside.

Expected Solution: This tool will use Machine Learning to help the user by proposition to assess the quality the following portfolio for a better risk management. It will save the time of the business to use it on the assessment evaluation.

Title: Resource efficiency indicator

Context: Currently, efficiency is measured through various means mainly involving trackers by the human resource department.

Description: We are looking for a technology or a solution to measure efficiency, productivity, utilization, time, which can provide data for operational decisions.

Expected Solution: AI/ML based tool to provide SMART, real time analysis and dashboards of utilization and forecast based on work allocated/pending.

Title: Text analysis to improve procurement efficiency

Context: Procurement team is sourcing thousands of “intellectual services” per year. Business needs and vendors proposals must follow a set of rules to be compliant with the regulation. Proof reading therefore becomes a mandatory and cumbersome process.

Description: Buyers and back office team read all Purchasing Requests to check the compliance with SG rules and bring some modifications like delete forbidden words, use of synonyms, change expressions and sentences, etc. We want to leverage on Natural Language Processing and Machine Learning to offer automated proofreading to ensure the Requests for proposals, business needs, and vendors’ proposals are compliant.

Expected Solution: A tool that will automatically proofread the business needs, request for proposals, and vendor proposals ensuring they are compliant with regulatory rules. We want to identify automatically noncompliant text within these documents and offer a compliant correction to the text. Via this automated process, we expect that business owners will become more independent in their procurement for smaller contracts. Same solution can be extended beyond “intellectual services” on a second step.

Title: Procurement performance analysis automation.

Context: The sourcing team is assessing its performance by evaluating achieved savings post negotiation via "Performance forms”, which are controlled to ensure they have right methodology of assessing the gains. We are willing to automate and enrich this manual process to make it systematic. We are looking for a solution to automatically validate some performance sheets and/or to assist the person in charge of validation. Challenge is not only about process automation, but also collection of figures, consistency check and calculation verification.

Description: Buyers within the Sourcing division fill in a “performance sheet” at the end of each project: amounts of covered perimeter and produced savings. All declared figures are calculated based on the “Performance methodology”, as SG-designed document describing various cases. The accuracy of declared savings, is manually verified by a controller. Control leads to validation (or rejection) of a performance sheet. Workload for performance controlling is currently huge. We want to automate this review. For Sourcing, this leads to a higher requirement in reporting reliability, hence to control more performance sheets. Our challenge is then to automate the control of a higher number of performance sheets.

Expected Solution: A tool on top of our sourcing solution checking the accuracy of each performance sheets posted by buyers. Format is common, languages is mostly French and English, requiring some NLP. Inputs from the startup on how to enrich the performance review are welcome (cost, delay, quality, other dimensions…). Procurement dataset available for the POC. Integration with the current procurement solution (and potential other sources in the long run) is a must. Error identification and automated correction proposals also welcome.

Title: Securities Services Net invoice alert

Context: In our Securities Services business, sub-custodian invoices are received every month and implemented into NetShare tool by the Net team based in Bangalore. The information about quantities and negotiated prices are partially available in NetShare. This tool automatically verifies invoice lines i.e. comparison between calculated theoretical invoice received and document from the sub-custodian. Team members can also review all invoices received every month to check that their global amount is coherent with the previous month.

Description: We want to automate and control all the invoice analytics - reviewing the variations from one month to another for all the lines - explain potential variation via the price calculation formula thanks to available KPIs and explainable global trends- generating a “coherence score” for the invoice- creating an “alert flag”, based both on the “coherence score” and the invoice amount - creating a monthly report on all sub-custodian invoices and an alert for “flagged” invoices. Data is easily accessible (no OCR required, no Citrix), but require an access to our specific tool. Artificial intelligence could be useful to determine rules.

Expected Solution: Robot with machine learning A first version of the tool able to review 5 to 10 sub-custodians invoices (2 years data) and identify the “alert flag” in case of incoherence in the invoice evolution. The explanation of the conclusion reached by the tool (why is the invoice flagged?)Analysis of the sub-custodian invoices, for the lines not already verified in NET Share Identification of an “coherence” score based on the evolution of the invoice since the previous month, not examined by KPIs changes or a “global trend” Global report (monthly) with flags on each invoice for review by bot or manually.

Title: e-retail compliance digitalization (partner challenge)

Context: The Compliance Operations team is a 1400 people team based out of India, managing compliances for product listing in ecommerce website for various English-speaking countries.

Description: The Compliance Operation team members manually check the compliance for every product with respect to certain set rules defined in the rules rationale and decide to list/delist the product on the website. On the basis of rules rationale, products are classified under four categories 1. Restricted product – e.g. Foot mat with Indian flag2. Dangerous Product– e.g. Knives, etc.3. HTS Trade – e.g. Shoes4. Food Safety Complaint – e.g. Chips Categorizing and checking the compliances for every product is time consuming, repetitive and requires lot of manual intervention. Hence, we want to revamp the current ways of working to be more automated, accurate and less effort intensive going forward.

Expected Solution: Streamlined operating model of Compliance Operation team with technology enablement:1. Constantly updating rules rationale with addition of SKUs and new products2. Need for specific Subject Matter Expertise for different/special categories, for e.g. pharma products3. Low success rate of current AI Engine (90% fall through rate)4. Linear growth of headcount to meet increasing volumes5. Managing surge in demand through capacity planning Solutions can address the following approaches: Crowdsourcing, Artificial Intelligence/Machine Learning, Gamification to incentivize behavior.