Банк внедрил AI-решения Deloitte для ускорения разработки и борьбы с мошенничеством

A financial services provider was ready to buy into the promise of Generative AI (GenAI). Ready to transform its software development life cycle (SDLC), fight back against increasingly sophisticated fraud attempts, and offer hyper-personalization to customers. But with the financial services provider’s critical data scattered and siloed, buying into GenAI felt like trying to make a purchase using coins buried between couch cushions and a wallet filled with multiple foreign currencies. Limited metadata documentation, unclear data definitions, restricted data searchability, and fragmented business intelligence environments each hindered the financial services provider’s ability to deliver on data requests quickly and obscured the lineage of critical data elements (CDEs), decreasing confidence in their quality. The financial services provider’s enterprise data and analytics organization grappled with these data governance and compliance challenges regularly, but particularly during quarterly risk and control self-assessment (RCSA) audits that demanded extensive manual reviews to piece together accurate data lineage to meet regulatory requirements. Introducing GenAI tools—which rely on timely, reliable, well-orchestrated data synthesized from all these disparate sources—would magnify the issues. Early explorations, like adding GenAI tools into software developers’ workflows, had hit a wall. The enterprise data and analytics organization and chief technology officer (CTO) suspected the lackluster results so far weren’t likely to improve without a major overhaul of data management and SDLC workflows. The financial service provider couldn’t afford to take too long addressing it either—competitors and bad actors were already using GenAI and wouldn’t slow down while it manually built data glossaries, traced lineages, and tested data quality for CDEs. WITHOUT STRONG DATA AND PROCESSES, THE PROMISE OF GENAI IS AN EMPTY ONE. The financial services provider found a solution in Data Assist™, part of Deloitte’s Ascend™ ecosystem. Deloitte worked with the enterprise data and analytics organization and CTO to design a Data Assist pilot with scrum teams in the bank fraud and property and casualty lines of business. Modular and vendor-agnostic, the suite of data solutions easily integrated into the financial services provider’s complex, multivendor infrastructure so that proprietary AI agents could expedite data discovery across the company’s fragmented sources, extract meaning from structured and unstructured sources to build metadata, and generate data governance artifacts. Data Assist’s Intelligent Business Glossary and Smart Data Quality features ingested thousands of disparate fields, documentation, and metadata, intelligently filling in gaps and resolving discrepancies to create a user-friendly business glossary with 5,000 enhanced definitions and 4,000 data quality measures for CDEs. The improvements in data literacy and discoverability were immediate. And catching. Inspired by the results of the initial pilots, other lines of business proactively brought their own use cases to the Deloitte team. Deloitte deployed additional existing Data Assist base capabilities—like Synthetic Data Generator to accelerate testing phases—and built new, custom capabilities—including a GenAI-based Data Model Reviewer to reduce manual review efforts—tailored to the company’s specific needs and to keep pace with ever-evolving external scientific research. As the data foundation grew stronger, leaders at the financial services provider were confident the right data environment was now in place and they could turn their focus back to their goals for augmenting the SDLC. The organization had equipped its developers with GenAI tools to automate certain coding tasks but had been disappointed to see lower adoption and less efficiency gains than expected. Deloitte helped assess the barriers to usage and discovered isolated tools on their own weren’t enough to meaningfully transform the SDLC. The financial services provider needed to provide more extensive education, establish change management processes, and integrate personas beyond just developers. Deloitte Ascend Agentic SDLC complemented the client's existing GenAI developer tools and brought new capabilities to its product owners and quality assurance engineers. In just four weeks, the team configured and installed platform plug-ins in Jira and team member local integrated development environments (IDEs) so the tool could support them right where they already worked. Product owners began using Ascend Agentic SDLC to generate user stories that pulled from past stories for context, and quality assurance engineers efficiently increased testing coverage with autogenerated integration tests. Rather than just offering isolated developer capabilities, GenAI now supported and unified a cohesive, augmented SDLC. Deloitte delivered extensive training in prompting and other skills to make sure each member of the organization’s scrum teams could make the most of the new tools. The strong data foundation built with Data Assist, comprehensive education and skills training, and more holistic ecosystem of users onboarded onto Ascend Agentic SDLC expanded the financial services provider’s GenAI potential beyond one-time efficiencies to new workflows and systems that could scale and transform the organization. GENAI PLATFORMS BRING POWER TO DATA AND SOFTWARE DELIVERY. PEOPLE AND PROCESSES GIVE THEM STAYING POWER. Throughout the pilots, scrum teams using Data Assist showed up to 80% reduction in time spent generating glossaries, critical data elements, and lineage. Enterprise data and analytics team members could focus on overseeing and validating the GenAI-created data governance artifacts, rather than slowly, manually generating them. By standardizing and accelerating the application of data governance to CDEs throughout the entire data life cycle, the organization reduced manual processing and friction to streamline the RCSA process and other compliance controls. Deloitte is now kicking off an engagement with the financial services provider to transform its entire data and analytics and business intelligence work through AI and agentic AI. Within the SDLC pilots, product owners and quality assurance engineers observed as much as 55% efficiency gains when leveraging Ascend Agentic SDLC. As these additional personas improved their processes and outputs, the low GenAI adoption among developers began to shift, with a more connected ecosystem of solutions reinforcing each persona’s contributions and compounding the team’s potential. To help maintain and expand this momentum, Deloitte provided a 12-month expansion plan covering team structure, financials, and hands-on training plans for thousands of product owners, developers, and quality assurance engineers. With a strong foundation and cohesive ecosystem of solutions across the data delivery and software development life cycles, the critical data that previously felt about as useful as spare coins in couch cushions has become a powerful resource for the financial services provider to drive real change.