Turning Legacy Tech into AI Engine
Key Points
- Legacy IT, often seen as a hindrance, actually houses the critical historical and real‑time data that fuels AI, so it should be viewed as an asset (“legendary”) rather than a burden.
- Breaking down data silos and integrating disparate systems—whether on‑premises, mainframe, or multiple public clouds—creates a unified environment essential for effective AI outcomes.
- An intentional hybrid‑cloud‑by‑design architecture allows AI workloads to operate wherever data and applications reside, delivering optimized analysis, built‑in governance, and cost‑effective data handling without unnecessary movement.
- Modernizing and unifying legacy technology into this hybrid framework unlocks “superhighway” pathways for innovation, enabling businesses to seize large‑scale opportunities and accelerate AI‑driven transformation.
Sections
- Turning Legacy Tech Into AI Engine - The speaker explains how existing legacy systems can be integrated into a hybrid‑cloud architecture to break data silos, leverage historical data, and power AI initiatives.
- Hybrid AI Architecture for Business Efficiency - The speaker stresses that building a stable, elastic, resilient, and secure hybrid AI system is essential for efficiently consolidating scattered data, cutting latency and costs, and delivering adaptive, high‑impact business insights.
- Hybrid Cloud AI for Legacy Systems - The speaker stresses that using smaller, tailor‑fit AI models within a hybrid‑cloud architecture lets organizations embed generative AI into existing mainframe and on‑premise workloads cost‑effectively, overcoming the misconception that full system rewrites are required.
Full Transcript
# Turning Legacy Tech into AI Engine **Source:** [https://www.youtube.com/watch?v=6-s_fUXP0FM](https://www.youtube.com/watch?v=6-s_fUXP0FM) **Duration:** 00:09:09 ## Summary - Legacy IT, often seen as a hindrance, actually houses the critical historical and real‑time data that fuels AI, so it should be viewed as an asset (“legendary”) rather than a burden. - Breaking down data silos and integrating disparate systems—whether on‑premises, mainframe, or multiple public clouds—creates a unified environment essential for effective AI outcomes. - An intentional hybrid‑cloud‑by‑design architecture allows AI workloads to operate wherever data and applications reside, delivering optimized analysis, built‑in governance, and cost‑effective data handling without unnecessary movement. - Modernizing and unifying legacy technology into this hybrid framework unlocks “superhighway” pathways for innovation, enabling businesses to seize large‑scale opportunities and accelerate AI‑driven transformation. ## Sections - [00:00:00](https://www.youtube.com/watch?v=6-s_fUXP0FM&t=0s) **Turning Legacy Tech Into AI Engine** - The speaker explains how existing legacy systems can be integrated into a hybrid‑cloud architecture to break data silos, leverage historical data, and power AI initiatives. - [00:03:14](https://www.youtube.com/watch?v=6-s_fUXP0FM&t=194s) **Hybrid AI Architecture for Business Efficiency** - The speaker stresses that building a stable, elastic, resilient, and secure hybrid AI system is essential for efficiently consolidating scattered data, cutting latency and costs, and delivering adaptive, high‑impact business insights. - [00:06:21](https://www.youtube.com/watch?v=6-s_fUXP0FM&t=381s) **Hybrid Cloud AI for Legacy Systems** - The speaker stresses that using smaller, tailor‑fit AI models within a hybrid‑cloud architecture lets organizations embed generative AI into existing mainframe and on‑premise workloads cost‑effectively, overcoming the misconception that full system rewrites are required. ## Full Transcript
[Music]
When you hear the term legacy technology,
...you might picture something old and outdated, like a crumbled road map that’s stuffed into your car’s console.
But in this episode, I’ll explain how your existing technology,
...as part of a hybrid-cloud-by-design architecture can be the engine that powers your AI strategy.
Your existing technology holds your data, and your data is the basis for your AI.
Welcome to AI Academy.
I’m Hillery Hunter, CTO of the IBM Infrastructure unit and General Manager of Innovation.
I work on the computers that you likely never see and ideally shouldn’t even know exist because they just work.
Let’s explore how to design a hybrid cloud architecture that works for and with your AI and your data.
Often, IT leaders think about their legacy technology, for example, older applications,
...mainframe and uncoordinated instances of public cloud usage as a hindrance to innovation.
It can be viewed as being not agile, as a drain on budgets and incapable of keeping up with the pace of change.
But AI doesn’t exist in a vacuum.
It gets it’s smarts from historical information, combined with current and instantaneous data to work effectively.
That’s why breaking down data silos and creating consistent environments across your IT estate,
...are key to realizing the true potential of AI.
Let’s go back to that map analogy that I started with.
Imagine you’re on a road trip, trying to get from point A to point B in your journey.
But imagine that there’s big ink blots all over the map,
...removing critical information and making it impossible to decipher the roads that connect you to your destination.
In this situation, you simply don’t have all the information to be effective and efficient.
Alternately, if your systems aren’t integrated, if your data isn’t integrated,
...if legacy IT isn’t modernized and brought to the table in this AI conversation, that’s like removing all of the interstates.
You no longer have the superhighway to get you where you need to be.
You’re stuck on the back roads and you can easily lose your way,
...because you’re not able to utilize the best technologies and all of your data.
You lose macro opportunities to solve big and bigger problems and get your business where it needs to go.
Simply put, don’t think about your existing tech as legacy,
...think about it as legendary, how each piece can help lead the AI journey to goal if it’s used properly.
In this AI era, it’s time to rethink your cloud strategies.
By choosing an intentional hybrid cloud architecture, you can bring your AI to wherever your data and applications already reside.
You’ll have the ability to analyze the data in an optimized, organized way,
...in the cloud and on premises, with governance built in and integrated across this landscape from day one.
Your data can be organized and governed without moving it, avoiding data movement costs and potential security compromises.
But if instead you don’t leverage an intentional hybrid approach, then your data is constantly being moved to where the AI is.
Training to your desired outcomes,
...tuning to your historical insights and applying AI to your current information takes time, money and work.
Moving all of this scattered data can become arduous and expensive very fast,
...like a GPS trying to decipher the best route in a constantly changing, underlying map.
Implementing an intentional hybrid architecture lets you tackle AI with confidence,
...consistency and enables you to course correct when needed.
The result is substantially different.
Latency improvements that can change business outcomes, meaning being able to get on and use the superhighway.
Cost efficiencies, meaning taking the most efficient route from point A to point B, and better insights.
This is then far beyond just having a paper road map.
This is providing your company with a GPS.
A system which is adaptable and understands the 3rd and 4th dimensions,
...like traffic and road conditions that don’t even exist on a 2D map.
Here are four considerations that come to my mind.
First, make sure your systems are stable and ready for AI.
The most important elements for system stability are elasticity, resiliency and security.
For elasticity, be sure your environments have appropriate capacity for expansion.
For resiliency, have high availability and disaster recovery in place, because AI applications will become essential to your business.
As you plan for AI to begin to operate on and leverage key data,
...this is an important time to check up on your cybersecurity management.
Next, adopt modern operations techniques.
When we talk about modernizing your operations,
...we’re talking about implementing platform engineering practices, formally DevOps, DevSecOps, etcetera.
Getting good at these will help the speed of AI creation and deployment.
Other crucial steps to modernizing your operations are to leverage infrastructure as code,
...compliance as code, modern application observability and automated workload rebalancing.
These steps will help you spend less on IT management and give you more time to focus on building smarter applications.
Be sure to provide access for the data.
While many enterprises have been through multiple rounds of data management efforts in the past,
...the time is truly now to put in place robust data cataloging and data governance systems.
These provide appropriate and seamless access for data, for creation of AI capabilities.
API-based modernization inserting points into an application where data and capabilities can be accessed is essential.
Lastly, optimize end-to-end deployment.
AI deployment is often feared to cause cost increases, but I want to put those fears to rest.
There are answers.
We’ve seen that automated workload optimization can help take out 30% or more of GPU usage,
...through management of system resources alone.
And use of smaller, tailor-fit AI models is key to cost management.
Most importantly, having hybrid cloud solutions that enable you to choose where your AI runs,
...means you can match up data with AI for a cost-efficient, latency-optimized solution.
What are the biggest roadblocks for technology leaders when it comes to AI?
First, it can certainly take some imagination to see that you don’t need to rewrite and rework everything.
That AI can be integrated into existing systems and workflows.
But it’s indeed very likely that your existing systems have powerful, secure and data rich capabilities,
...that can give you a competitive edge.
Today, AI can be deployed to great effect, even within a mainframe-hosted application.
It can be deployed on premises, on systems hosting your critical data, in the cloud or at the edge.
When you power your enterprise with hybrid cloud capabilities,
...tailor-fit models, governance and best practices, you can drive real value with AI.
Implementing intentional hybrid cloud architecture is a business-critical decision,
...that ensures your systems are well protected, streamlined and optimized.
Every new journey has its risk, but if you don’t go, you’ll never grow.
In IBM’s CEO Arvind Krishna’s 2024 Think keynote,
...he gave us a preview of a new class of generative AI assistants for advising and helping users of the mainframe.
Helping the IT operator to do their job better and making these systems smarter.
Across the enterprise IT landscape, new AI assistants will range from helping coders to transform enterprise applications;
...to hybrid cloud assistants that facilitate system optimization;
...to code explainers that help developers understand and document applications through natural language.
These are some of the many innovations and accelerations your company can take advantage of,
...by adopting a hybrid-cloud-by-design architecture for your AI.
And the time is now.
Being able to get from point A to point B is the beginning of any road trip.
So is making the most out of your existing IT resources to get the most from your AI.
Whether on well-traveled routes or lands beyond your imagination,
...with the right tools and the ability to see the road ahead with clarity and confidence, it’s going to be a legendary journey.
Ready? Let’s go.
[Music]