At InterSystems READY 2025, a recurring message resonated across sessions: the most significant barriers in supply chains today are not futuristic, nor are they rooted in the complexity of AI models. Instead, they lie in the foundational issues of fragmented, inconsistent, and unreliable data.

The session “Solving Supply Chain Challenges with Data, Driven Intelligence” focused on the practical steps organizations must take to unlock the value of supply chain data. The discussion was led by Mark Holmes – Head of Supply Chain Market Strategy, Ming Zhou – Head of Supply Chain Product Strategy and Emily Cohen – Senior Solution Developer. Together, they mapped out the realities of supply chain data challenges and presented approaches that are less about grand visions and more about achievable steps: reconcile the data, automate repetitive work, and then apply intelligence in a way that improves day, to, day performance.

Why Supply Chain Data Remains a Bottleneck

Supply chains have become increasingly digitized, but digitization has not solved the core issue of data fragmentation. Procurement teams often operate with supplier records scattered across multiple ERPs. Logistics departments rely on siloed warehouse management systems. Planning teams pull reports from disconnected forecasting applications.

Mark Holmes pointed out that this patchwork of systems leads to duplicated supplier records, mismatched product identifiers, and time lost reconciling basic facts. These are not rare occurrences but daily realities. The consequence is predictable: planning decisions are made on flawed inputs, delays cascade through the network, and advanced analytics projects fail before they begin.

Ming Zhou added that while many organizations rush toward predictive AI, the truth is that most forecasting models fail because they are built on weak data foundations. Without consistency, even the best model produces unreliable outputs.

Emily Cohen emphasized that this is where organizations need to focus first, not on sophisticated models, but on establishing a baseline of clean, validated, and governed data.

Data Fabric Studio: A Practical Toolset

The centerpiece of the discussion was InterSystems Data Fabric Studio, a platform designed to connect disparate data sources, Snowflake, Kafka, AWS S3, and ERP databases, and transform them into unified, reliable datasets.

Unlike traditional ETL (Extract, Transform, and Load) projects that require months of coding and testing, Data Fabric Studio employs recipes, configurable workflows that clean, reconcile, and standardize data. These recipes automate repeatable processes, ensuring that once supplier records are aligned or product codes are standardized, the consistency holds over time and applied to add data sets across data sources.

Mark Holmes explained that this approach eliminates the cycle of one, off data projects that fall apart as soon as new data flows in. Instead, organizations can lock in data quality improvements and free staff from repetitive, manual reconciliation.

Case Study: Supplier Data Across ERPs

One example shared by Holmes and Cohen involved supplier records managed across two ERP systems. The inconsistencies were predictable but damaging:

One supplier might appear under multiple names.
Different identifiers were used across systems, complicating invoice matching.
Purchase orders could not be reconciled without manual intervention.

By applying Data Fabric Studio, the team:

Mapped suppliers to a single source of truth using identifiers such as DUNS numbers.
Standardized supplier names and records across systems.
Built lookup tables to automatically reconcile discrepancies in the future.
Scheduled daily refreshes so data quality stayed intact.

The result was a cleaner supplier database, faster onboarding, and fewer invoice disputes. What stands out in this example is not the sophistication of the solution but its practicality. The gains came from structured data reconciliation, not from exotic algorithms.

Forecasting Through Structured Snapshots

Zhou shifted the focus to forecasting. His point was simple: forecasts are only as good as the data used to build them. Too often, planners must run ad hoc queries across inconsistent systems, leading to variable inputs and unstable forecasts.

The recommended practice is to create structured data snapshots, capturing consistent baselines such as:

Open purchase orders every Monday morning.
Inventory by location at shift change.
Fulfillment cycle times at the close of each reporting period.

These snapshots provide planners with stable, repeatable inputs. While this may sound basic, the effect is significant: forecasting accuracy improves because the inputs are reliable, and planners spend less time chasing down missing data.

Zhou was clear that this is not advanced predictive AI. Instead, it is the groundwork that enables predictive AI to succeed. Without clean, consistent snapshots, AI models are destined to fail.

AI, Ready Data: From Vector Search to RAG

Cohen emphasized that AI does not fail because of weak models, it fails because of bad data. Large language models, predictive algorithms, and advanced optimization engines all require structured, validated, and governed data. Without it, the insights generated are misleading at best and damaging at worst.

To address this, Data Fabric Studio incorporates tools for vector search and retrieval, augmented generation (RAG). These enable:

Semantic search across suppliers, contracts, or parts databases, allowing staff to locate the right information even when queries are imprecise.
Feeding current and validated data into language models so that natural language queries return fact, based answers.
Allowing non, technical staff to use natural language interfaces that generate SQL queries or summarize trends.

Prescriptive Insights: Non, Traditional Data as Signals

Holmes expanded the conversation by drawing an analogy from the healthcare sector. In a study presented earlier this week, researchers found that analyzing patients’ shopping habits, specifically purchases of over, the, counter medication, could reveal early indicators of ovarian cancer before any clinical diagnosis was made.

This insight is directly applicable to supply chain management: valuable signals may not always be derived from conventional dashboards. Anomalies in supplier invoices, discrepancies in delivery documentation, or shifts in employee communications could help identify emerging risks before they are detected through traditional metrics. Organizations that systematically integrate these non, traditional data sources into their analytics framework are better positioned to identify disruptions at an earlier stage.

A central theme involves prescriptive insights enabled by AI, ready data. For example, to prevent procedure cancellations, such as a heart surgery being postponed due to a missing valve kit component, the application of advanced, AI, driven prescriptive analytics is critical. As demonstrated by Ming in his presentation, predictive tools identified which surgeries were at risk of delay or cancellation due to unavailable inventory. By leveraging AI, enabled insights, the team proactively sourced the missing components from another warehouse, ensuring surgical schedules remained intact. This outcome underscores the importance of not only preparing data for AI but also implementing advanced supply chain optimization through intelligent prescriptive solutions.

Modular Deployment: Start Small, Scale Gradually

A recurring point from Zhou was the importance of modularity. Data Fabric Studio does not require wholesale system replacement. Organizations can begin with a single use case, supplier data reconciliation, for example, and expand gradually to include forecasting snapshots, vector search, or natural language assistants.

This modular approach minimizes risk and allows organizations to demonstrate value incrementally. It also makes it easier to integrate with existing ERP, warehouse management, and planning systems rather than replacing them outright.

Scalability and Infrastructure

Finally, the speakers emphasized scalability. InterSystems IRIS, the engine behind Data Fabric Studio, has already been proven in healthcare environments, where it supports hundreds of millions of real, time transactions.

For supply chains, this track record matters. As data becomes central to operations, the infrastructure must scale without becoming a bottleneck. Inconsistent or unreliable infrastructure undermines even the best data practices.

Key Takeaways

From the READY 2025 session, the roadmap outlined by Holmes, Zhou, and Cohen is clear:

Reconcile and harmonize data across systems. Clean data is the foundation of everything that follows.
Automate repetitive processes. Recipes in Data Fabric Studio reduce manual reconciliation and enforce consistency.
Use structured snapshots for forecasting. Reliable baselines are essential for both planners and predictive AI.
Introduce AI gradually. Take care of data first, and then apply the right AI technology one use case at a time, and grow from there.
Ensure infrastructure scalability. Proven engines like InterSystems IRIS reduce risk as volumes grow.

A Disciplined Order of Operations

The session leaders were clear: digital transformation in supply chains is not about chasing the latest technology. It is about establishing discipline in the order of operations:

Get the data right.
Automate manual tasks.
Scale the infrastructure.
Apply AI only when the groundwork is complete.

This sequence ensures that AI enhances decision, making rather than amplifying bad data.

Intersystems READY 2025 event, and especially the session “Solving Supply Chain Challenges with Data, Driven Intelligence” underscored that the most effective supply chain strategies are practical, not speculative. By focusing first on unifying and governing data, organizations can lay the foundation for automation, forecasting, and AI applications that deliver real value.

The lesson is straightforward but often overlooked: data comes first, intelligence comes later. Supply chains that adopt this discipline will not only resolve today’s data bottlenecks but also position themselves to adapt to the demands of tomorrow’s networks.

The post Solving Supply Chain Challenges with Data-Driven Intelligence – Practical Steps to Unlock the Value of Supply Chain Data appeared first on Logistics Viewpoints.