View Full Transcript
WEBVTT
00:00:00.000 -> 00:00:07.000
Narrator: We will show you an agentic AI demo about autonomous predictive maintenance for manufacturers.
00:00:07.000 -> 00:00:31.000
Narrator: Enterprises are moving fast on agentic AI, but progress often stalls because critical data is spread across multiple clouds, regions, and on-prem systems, and can't be replicated due to regulatory, sovereignty, or cost constraints. Agentic AI needs live, trusted, contextual data, and without a unified data foundation, organizations struggle to scale impact.
00:00:31.000 -> 00:00:49.000
Narrator: Together, Denodo and AWS address these challenges by delivering a comprehensive, open architecture that enables organizations to leverage their entire data estate across multi-cloud, hybrid, and on-premises environments while ensuring compliance and security.
00:00:49.000 -> 00:01:04.000
Narrator: The Denodo Platform acts as the logical data access layer that simplifies this distributed data, making your entire AWS data ecosystem and the rest of your enterprise information instantly available, governed, and truly AI-ready.
00:01:04.000 -> 00:01:49.000
Narrator: Denodo and AWS help industrial organizations build and deploy AI and analytics solutions that bridge the gap between the factory floor and the enterprise. In this demo, we'll walk you through our joint solution capabilities using an example from industrial asset performance management. Imagine you're a global manufacturing firm that has built and deployed AI agents on Amazon Bedrock and Amazon SageMaker AI to optimize your production uptime and equipment maintenance. Current manufacturing and maintenance data is hindered by system fragmentation where critical data is trapped in disconnected systems of record. This leads to problems such as unplanned downtime, decreased overall equipment effectiveness, and high operational costs caused by reactive rather than predictive maintenance.
00:01:49.000 -> 00:02:42.000
Narrator: Instead of manual maintenance audits and reactive repairs, this solution deploys a multi-agent system that acts as an autonomous reliability engineer fueled by a real-time data fabric. Denodo, the sensory nervous system, acts as a logical data fabric that creates a unified, real-time view of the industrial asset. It joins live machine telemetry from AVEVA PI with on-premises ERP inventory, maintenance logs, and golden run historical baselines. This allows AI agents to sense the factory floor without complex data movement. AWS, the cognitive engine, and Amazon Bedrock provide the reasoning capabilities. These agents consume the governed data products from Denodo to perform high-fidelity anomaly detection, root cause assessment, and automated maintenance orchestration.
00:02:42.000 -> 00:03:02.000
Narrator: The result: a transition from reactive, manual maintenance oversight to autonomous asset orchestration. By providing a live data foundation, the system enables straight-through maintenance for standard mechanical issues. AI agents autonomously handle the detection, assessment, and scheduling of repairs, allowing human collaborators to focus only on the most complex exceptions and high-value equipment cases.
00:03:02.000 -> 00:04:13.000
Narrator: Let's walk through a real-world maintenance failure scenario for FiberTech Materials, the global leader in industrial components. Currently, their machine producing carbon fiber placement is exhibiting high heat signatures. This machine is a critical bottleneck for the entire facility. If it fails unexpectedly, the consequences are severe, such as: a high-value carbon fiber panel may be ruined, putting a $78,000 financial risk; a critical delivery deadline for a high-value partner is missed; the unplanned downtime triggers an additional contract penalty and halts downstream production. Denodo integrates with the on-premises SAP for materials and asset management and the machine telemetry data streamed from AVEVA PI. Denodo also integrates with ServiceNow for field service data and Amazon SageMaker Lakehouse for manufacturing execution data and lifecycle management information. Amazon Bedrock Agent acts as the autonomous predictive maintenance supervisor, finding root cause of potential failures, determining the best technician to repair the part, ensuring the best technicians are assigned to high-impact issues, and verifying the repair through telemetry data.
00:04:13.000 -> 00:04:54.000
Narrator: [UI showing Amazon Quick Automations workflow] Here is an Amazon Quick automation for autonomous predictive maintenance for manufacturing. Each step in the automation includes an integration with Denodo to provide the agentic AI-ready data for each step of the process. Here you can see agentic orchestration for the predictive maintenance flow.
00:04:54.000 -> 00:05:13.000
Narrator: The AI agent constantly monitors machine health. It uses real-time and historical data from Denodo logical data products to perform triage. For example, it identifies the specific asset at risk and calculates the potential financial impact, such as a ruined carbon fiber panel or a contract penalty.
00:05:13.000 -> 00:05:30.000
Narrator: Denodo acts as the foundational access layer. The diagnostic agent uses Denodo's MCP integration to instantly query machine, factory, and sensor telemetry that is otherwise sprawled across complex on-premises and cloud environments.
00:05:30.000 -> 00:05:53.000
Narrator: [UI displaying Amazon SageMaker notebook] A key benefit of Denodo's unified semantic layer is reusability. The exact same business context and definitions used by AI agents are available to all other data consumers. For example, here is an AWS SageMaker notebook consuming the view from the first step.
00:05:53.000 -> 00:06:28.000
Narrator: [UI showing Denodo Data Marketplace] Denodo's Data Marketplace enables business users and data consumers, such as AI agents, to leverage the logical data products. The Denodo Platform is a complete no-code, AI-enabled graphical interface to create these logical data products. Here we are reviewing already created logical data products. Each of these products is used in the predictive maintenance workflow. Here we will focus on the predictive failure analysis logical view used in the root cause analysis agent.
00:06:28.000 -> 00:06:44.000
Narrator: [UI highlighting Data Lineage feature] Here you can see this view is integrating telemetry data on-premises, production lifecycle data in Redshift, and field service data in ServiceNow. The AI agent queries this data from Denodo, and Denodo provides zero-copy access to this data in a logical view.
00:06:44.000 -> 00:07:06.000
Narrator: The AI agents are able to make trusted, informed decisions, not only because Denodo is providing real-time access to the data, but also because it provides the agent with rich semantic metadata. Denodo acts as a unified semantic layer by automatically synchronizing with the Amazon SageMaker Catalog. This ensures that the exact same trusted business definitions used by your human data scientists are seamlessly fed to your AI agents.
00:07:06.000 -> 00:08:02.000
Narrator: [UI displaying Global Security Policies in Denodo Design Studio] Denodo enables you to utilize centralized policies, including granular attribute-based access control, dynamic data masking, and end-to-end data lineage to govern all data at once, from on-premises systems to cloud data services and back, ensuring your users and AI agents have access to the right data while maintaining strict regulatory compliance and building enterprise trust. Shown here is a global security policy implementing a masking policy based on views and columns that are tagged with PII insurance.
00:08:02.000 -> 00:08:33.000
Narrator: With Denodo, the AI agent runs real-time analysis on both the live machine telemetry and historical maintenance and production systems data stored across different clouds and on-premises environments. It is able to predict likely outcomes, such as an imminent bearing failure, and determines which anomalies require a human in the loop for a floor supervisor for a complex repair versus those that can go to straight-through maintenance for autonomous scheduling and technician resolution. Denodo provides a unified semantic layer that connects data across AWS data services alongside your non-AWS sources, such as SAP and AVEVA PI. Denodo ensures all consuming applications, from AI agents to BI tools, have the right business context to generate accurate insights and recommend and execute the right actions.
00:08:33.000 -> 00:08:58.000
Narrator: The diagnostic agent completes its analysis and identifies the root cause: a bearing failure driven by excessive vibration and heat. By cross-referencing historical maintenance logs, the AI spots an 85% match to a recurring wear pattern from earlier this year, likely due to inadequate lubrication. Most importantly, it calculates that there are just 18 hours until a total breakdown. Because standard procurement takes 14 days, the AI immediately flags this as a critical emergency with revenue directly at risk.
00:08:58.000 -> 00:09:12.000
Narrator: The maintenance parts readiness agent determines there is revenue at risk. The agentic AI immediately justifies the $150 part replacement. It authorizes the release of the staged inventory and triggers an emergency technician dispatch. To keep the safety stock secure, the agent also autonomously generates replenishment purchase orders with our primary and backup suppliers and updates the maintenance schedule to prevent this from happening again.
00:09:12.000 -> 00:09:38.000
Narrator: With the financial and logistical decisions made, the dispatch agent takes action. It instantly identifies a 15-minute production gap in the maintenance schedule and cross-references ServiceNow to assign our lead technician, Alistair Johnson, for an immediate repair.
00:09:38.000 -> 00:10:06.000
Narrator: In this final step, the operations agent performs a closed-loop verification to confirm the repair was successful. It actively pulls the latest telemetry, validating that both vibration and temperature have safely returned to their baseline thresholds. With the data stabilized, the AI officially marks the machine as healthy, closes out the incident tickets, and updates the predictive maintenance baselines for the entire fleet to prevent this specific failure from happening again.
00:10:06.000 -> 00:10:17.000
Narrator: [UI displaying summary slides] By building and running their agentic AI solutions on AWS and Denodo, FiberTech Materials reduces operational waste, specifically through automated thermal threshold verification and predictive failure detection, directly lowering unplanned downtime costs and providing a multi-million dollar impact on production yield. This enables an increase in manufacturing throughput without increasing headcount by transitioning low-complexity maintenance tasks to straight-through maintenance.
00:10:17.000 -> 00:10:28.000
Narrator: Unified Data. Powering Agentic AI at Scale. Learn more at denodo.com/aws. Thank you.
Learn how a logical data fabric powers Agentic AI to automate predictive maintenance on the manufacturing floor. Connect live machinery telemetry to ERP data in real time to prevent downtime.
Deploying operational AI on the manufacturing floor usually hits a major roadblock: data silos. To predict a machine failure, order a replacement part, and schedule a technician, an AI agent needs data from completely different worlds—live factory sensors, supplier inventory databases, and corporate financial ledgers. Moving all this data into one place takes too long and introduces costly delays.
This quick walkthrough with Denodo and AWS shows a better approach. By creating a unified logical data fabric, the system delivers live operational context from separate data sources exactly when it is needed, without copying or moving the underlying files. Watch how an autonomous AI agent instantly detects a machine anomaly, identifies the root cause, checks regional part availability, runs a financial impact analysis, and schedules the right technician—turning complex operational data into an automated business decision.
What you'll learn:
- How to feed live factory floor telemetry directly into an enterprise AI engine.
- The business value of connecting operational tech (OT) with corporate IT systems without slow ETL pipelines.
- How AI agents use a logical data fabric to securely check supply chains and vendor data.
- Automating multi-step workflows—from anomaly detection to final dispatch—using trusted data.
