Retail and commercial banks are at a pivotal moment. The pressure to deliver smarter, faster, and more accurate decisions is intensifying, driven by rising customer expectations, competitive threats, increasingly sophisticated fraud, and regulatory scrutiny. At the heart of this transformation lies decision intelligence—the ability to harness data and AI to make better business decisions at scale and often in real time.
According to Celent’s report“It’s All About the Data!”, banks are prioritizing investments to improve the performance of decision intelligence models for fraud detection, credit risk modeling, customer engagement, and loan and account origination. These use cases are powered by AI, but their success hinges on a critical factor: AI-ready data.
This blog post explores how retail banks can meet the mandate for decision intelligence by building a foundation of AI-ready data, supported by insights from the Gartner® report, “A Journey Guide to Deliver AI Success Through AI-Ready Data.” It also outlines how a data fabric architecture can help banks operationalize AI-ready data and offers best practices for implementing this architecture in a non-disruptive way.
The Decision Intelligence Mandate in Banking
Celent’s research highlights a clear trend: retail banks are investing in improving the models and data infrastructure powering AI-driven decision intelligence for risk management, customer experience, and to drive better operational efficiency. However, their investments in data platforms continue to hold back model performance due to fragmented, siloed, and poor-quality data.
“Despite all the investments banks have made in data platforms, data issues remain pervasive,” Celent notes. “Decision intelligence hinges on the ability to harness data effectively.” – Celent: “ It’s All About The Data!” |
This disconnect between ambition and execution is a major barrier. AI models—especially those used in GenAI applications—require data that is not only clean and accessible but also contextually relevant, representative, and governed. This is where the concept of AI-ready data becomes essential.
AI-Driven Decision Intelligence Requires AI-Ready Data
According to Gartner® “AI-ready data can only be determined contextually to the use case and AI technique, which requires rethinking data management.” In essence, AI-ready data as data that is fit for purpose—aligned to the specific AI use case, and governed to ensure trust and compliance. Unlike traditional data management, which emphasizes uniformity and cleansing, AI-ready data embraces complexity, including outliers and anomalies that are crucial for training robust models.
“Understand what AI-ready data is and what it means for your specific AI use cases. Remember, not all AI use cases require the same kind of data, nor do they need the same kind of data management support and stringent SLAs.”
– Gartner: “A Journey Guide to Deliver AI Success Through AI-Ready Data” |
This insight is especially relevant in retail banking, where each decision intelligence use case demands a different mix of data types, sources, and governance models. For example, fraud detection may rely heavily on transactional and behavioral data, while customer engagement may require integrating structured CRM data with unstructured call center transcripts and social media signals.
To meet these diverse needs, banks must adopt a multimodal data strategy—one that supports structured, unstructured, and streaming data, and enables semantic enrichment and contextualization. This is the essence of AI-ready data: data that is not only accessible and clean, but also contextually aligned, governed, and fit for purpose.
Data Fabric: The Architecture for AI-Ready Data
Both Gartner and Celent identify data fabrics as a key enabler of AI-ready data. A data fabric architecture provides intelligent, unified access to distributed data across silos, systems, and formats. It supports real-time integration, semantic enrichment, and governance—making it ideal for AI-driven decision intelligence.
| “The next generation of the logical data warehouse pattern is the data fabric pattern,” Gartner notes. “Ideal for organizations that need to take their data architecture to the next level.” – Gartner: “A Journey Guide to Deliver AI Success Through AI-Ready Data” |
Celent adds that a data fabric can help banks overcome persistent data challenges by enabling:
- Real-time data access for decision-making
- Metadata-driven governance for compliance
- Multimodal data integration for GenAI use cases
- Scalable orchestration for AI model deployment
In short, a data fabric transforms data chaos into data confidence—allowing banks to deliver AI-powered decisions with speed, accuracy, and trust.
Best Practices for Implementing a Data Fabric in Banking
Implementing a data fabric is not just a technology upgrade—it’s a strategic enabler for delivering AI-ready data at scale. According to Gartner, “[i]nvest in data fabric to allow your data management infrastructure to be resilient.” A successful data fabric implementation requires a blend of foundational data management, AI-specific innovations, and iterative scaling. For retail and commercial banks, this means building a resilient, intelligent data architecture that supports real-time decision intelligence across use cases.
Here are six best practices for implementing a data fabric, grounded in Gartner’s guidance and aligned with the needs of modern banking:
1. Build on Existing Data Management Foundations
Rather than replacing legacy systems, extend your current data architecture with data fabric capabilities. Gartner emphasizes that foundational components like data integration, metadata management, and data governance remain essential.
Why it matters: Retail banks already have significant investments in data warehouses, data lakes, and operational systems. A data fabric should augment—not disrupt—these investments, enabling faster time to value.
2. Incorporate AI-Specific Data Innovations
To support GenAI and advanced analytics, banks must go beyond traditional data tools. Gartner recommends integrating innovations such as:
- Vector databases for semantic search and retrieval-augmented generation (RAG)
- Chunking and embedding for unstructured data
- RAG orchestration to enrich LLMs with enterprise context
Why it matters: These capabilities are essential for powering AI-driven decision intelligence use cases like fraud detection, credit risk, and customer intelligence.
3. Invest in Metadata, Lineage, and Observability
A data fabric must be intelligent. That means embedding active metadata management, data lineage, and observability into the architecture to ensure transparency, trust, and performance monitoring.
Why it matters: In regulated environments like banking, explainability and auditability are non-negotiable. These capabilities also support faster troubleshooting and model validation.
4. Support Multistructured and Multimodal Data
Retail banking use cases increasingly rely on a mix of structured (e.g., transactions), semi-structured (e.g., JSON), and unstructured (e.g., emails, PDFs) data. Gartner stresses the importance of supporting multistructured data to enable GenAI and other advanced AI techniques.
Why it matters: A data fabric that can unify and contextualize multimodal data enables richer insights and more accurate AI models—especially for customer intelligence and compliance use cases.
5. Enable Composable, Scalable Architectures
Gartner recommends adopting a composable architecture that allows banks to incrementally add new capabilities—such as RAG, knowledge graphs, or FinOps—without rearchitecting the entire system.
Why it matters: This modular approach supports agile innovation and reduces risk. Banks can start with a single use case (e.g., real-time credit scoring) and scale to others as business needs evolve.
6. Embed Governance and AI Readiness into the Fabric
Governance is not an afterthought—it must be embedded into the data fabric from day one. This includes role-based access, data quality controls, and alignment with AI governance frameworks.
Why it matters: As GenAI becomes more pervasive, banks must ensure that only approved, high-quality data is used in AI models. This protects against bias and regulatory exposure.
InterSystems Smart Data Fabric: Built for Scalable, AI-Ready Data in Banking
InterSystems solutions are purpose-built to meet the need for an agile, intelligent, and resilient data architecture that AI-driven decision intelligence demands. Our smart data fabric approach delivers real-time, governed access to distributed data while supporting the full lifecycle of AI and GenAI use cases.
Here’s how InterSystems aligns with Gartner’s best practices for implementing a data fabric:
1. Non-disruptive data layer
InterSystems solutions seamlessly integrate data from existing data warehouses, lakes, and operational systems—without requiring data replication or replatforming. We enable on-demand access to live data across silos, preserving prior investments while modernizing the data architecture.
2. Embedded AI and ML Capabilities
InterSystems includes native support for vector search, retrieval-augmented generation (RAG), and embedding techniques—allowing banks to enrich GenAI models with enterprise context and embed models into real-time operational workflows. These capabilities are embedded directly into the platform, reducing complexity and accelerating deployment.
3. Process Orchestration
InterSystems enables process orchestration by integrating internal and external data and services, AI model outputs, and business workflows into a unified architecture. Banks can trigger real-time decisions—such as fraud alerts or credit approvals—based on live data and model outputs. This orchestration layer ensures that insights are not just generated, but acted upon quickly and contextually, embedding intelligence directly into operational processes.
4. Metadata Management, Lineage, and Observability
The platform provides metadata management, data lineage tracking, and real-time observability. These features ensure transparency, trust, and performance monitoring—critical for regulated banking environments and for maintaining model integrity over time.
5. Multistructured and Multimodal Data
InterSystems supports structured, semi-structured, and unstructured data, including documents, emails, and non-traditional data sources. This multimodal capability enables richer insights and supports advanced use cases like fraud detection, credit analysis, and risk management.
6. Composable and Scalable Architectures
With its phased implementation approach, InterSystems allows banks to start with a single use case—such as fraud detection or credit risk—and scale incrementally. New capabilities can be added without disrupting existing systems, supporting agile innovation and continuous improvement.
7. Governance and AI Readiness
Governance is built into the core of InterSystems solutions. Features like role-based access controls, data quality monitoring, and compliance auditing ensure that AI models are trained and operated on trusted, approved data—reducing risk and supporting responsible AI practices.
By implementing InterSystems smart data fabric, retail and commercial banks can accelerate their journey to AI-ready data, unlock the full potential of decision intelligence, and future-proof their data architecture for GenAI and beyond.
Conclusion: Building the Foundation for AI Success
Retail and commercial banks are under pressure to deliver smarter decisions, faster. Data availability, quality, and timeliness are the biggest drivers of model performance, whether for decision intelligence systems or advanced analytics leveraging GenAI. As Celent and Gartner both emphasize, AI-ready data is the foundation for scalable, responsible AI.
By adopting a data fabric architecture, banks can unify their data, enrich it with context, and govern it with confidence—unlocking the full potential of AI across the enterprise.
Now is the time to act. Assess your data readiness, align your architecture to your AI ambition, and invest in the tools and practices that will make your data truly AI-ready.
Download the Celent whitepaper
Gartner, "A Journey Guide to Delivering AI Success Through ‘AI-Ready’ Data," Ehtisham Zaidi, Roxane Edjlali, 18 October 2024. GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally and is used herein with permission. All rights reserved.
Celent: “It’s All About The Data!” Colin Kerr and Craig Focardi, 01 May 2025.
