Businesses are losing out twice by having to support old IT systems that they cannot afford to turn off because they perform business-critical functions.
Companies have to pay to keep computer systems running well beyond their natural life, which means they have fewer resources to invest in new technology.
“They are spending money improving customer experience and improving efficiencies in the business, but the legacy makes all of those projects harder,” says Don Schuerman, chief technology officer (CTO) at Pegasystems.
“It limits the ability of organisations to build the cycle of innovation and continuous change that is essential to innovation,” he tells Computer Weekly.
Pegasystems, a $140bn turnover technology company based in Waltham, Massachusetts, develops the software used behind the scenes at some of the world’s largest organisations to manage and automate their business processes.
A survey of more than 500 IT decision-makers, commissioned by Pega, shows that half of the companies questioned rely on computer systems that are over 10 years old, and one in 10 have at least one computer system that is between 20 and 30 years old.
Businesses need courage
Companies need to be more courageous in tackling the problem of ageing IT, according to Schuerman, with old technology becoming increasingly difficult to support over time.
“There are many systems where the teams that built them are not around anymore. The people who understand the code base are no longer there,” he says.
Suppliers like Pega are developing tools that make it easier for organisations to reduce their reliance on outdated IT systems. Both Amazon Web Services (AWS) and Accenture, for example, have developed transformation tools that use artificial intelligence (AI) agents to help companies migrate their workloads from mainframe computers to datacentres.
Pegasystems has expanded its own tool, Blueprint – a rapid prototyping tool for creating business workflows – with agentic AI.
“Companies are spending money improving customer experience and improving efficiencies in the business, but the legacy makes all of those projects harder”
Don Schuerman, Pega
It uses AI agents to ingest and analyse videos of legacy software, software documentation, technical files, screenshots and source code, or a company’s website, to create blueprints that can be turned into cloud applications.
Pega demonstrated its capabilities at the company’s conference this week. The company showed how Blueprint was able to replicate the functions of a credit card management system originally written in Cobol, a computer language that dates back to 1959.
AI agents built into Blueprint were able to interrogate a video showing someone describing how they use the software, along with software documentation.
Blueprint was able to identify the functions of the code, replicate its workflows in a matter of minutes, and make suggestions on how to improve it.
A live demonstration showed how, with the addition of an AI bot, a customer could phone their bank, hold a natural language conversation with an AI agent, check their credit card balance, request an increase in their credit limit and have it approved, without human intervention.
Another example is Vodafone, which is using Blueprint to develop applications across its business. The company is developing prototypes that it can share and test with customers and people in the business before the apps are deployed.
The phone company says using Blueprint has allowed it to speed up the development and deployment of applications, with one application going from concept to production in just 48 hours.
The real value of Blueprint is that it changes the conversation from a “lift and shift” mentality, says Schuerman, where organisations typically take legacy software and repackage it in a Kubernetes container or equivalent to run in the cloud.
Now, organisations can focus on improving their business processes rather than duplicating them.
Blueprint can complete 50% to 60% of the development work of an app, but it will never replace developers, says Schuerman.
Companies will still need to take data used by legacy applications and migrate it to cloud storage. Google, Pega, AWS and others have developed tools for that task.
Blueprint cannot replace all legacy technology, says Schuerman.
“The stuff we are going to be moving and transforming is still going to be things that look naturally like good Pega apps. So, things that have workflow, things that drive degrees of process orchestration,” he says.
Blueprint would not be suitable for replacing financial ledger systems, SAP and other enterprise resource planning (ERP) systems, for instance. That said, Schuerman says Pega has had interest from companies that want to move the customised processes surrounding their ERP systems into Pega.
There is nothing, in theory, to stop organisations using Blueprint to implement work processes in other technologies outside of Pega.
The software is able to produce documents describing detailed workflows, the underlying data models and the interface points of the process.
“If you wanted to take a document and treat it as a requirements document, I think it would be a pretty good requirements document for traditional software development,” he says.
There are always going to be applications that can and should operate outside Pega. Security is one example, such as managing the authentication and authorisation of users and privileges.
Keeping the agents in check
One of the risks of using AI is that it can be unpredictable and prone to hallucinations. Pega solves this problem by using its AI agents to design and develop repeatable workflows, rather than allowing them to make unscripted decisions when running live.
In the case of a credit card provider, for example, AI bots can interact with customers to select the appropriate workflow, but do not have the power to make “runtime” decisions for the customer.
We are working to make sure you can wrap and plug agentic AI into a workflow that has a high degree of predictability and repeatability Don Schuerman, Pega
No bank is going to want to use a large language model to decide credit limits for customers when there is a risk that the model would make different recommendations for different customers without being able to explain how or why it made the decision.
“We are working to make sure you can wrap and plug agentic AI into a workflow that has a high degree of predictability and repeatability, and other people are not doing that,” says Schuerman.
In the past, Pega advocated the idea of “wrap and renew”, which entailed putting a software wrapper around mainframes and other legacy technologies to integrate them into other parts of the business. Now, it’s a case of “rethink and replace”, according to Schuerman.
Pega has plans for a new service that will make it possible for businesses to manage agentic AI, both those running on Pega’s software and those outside Pega, to ensure they are working in a coordinated way, which it calls Pega Agentic Process Fabric.
The service, which will be available from this autumn, will audit and register AI agents, workflows and data used on Pega and other IT systems. It will also be able to generate new workflows on the fly, with human oversight, if the process does not already exist.
The idea is to make sure that AI agents performing different tasks work together rather than in a conflicting way.
Can legacy technology ever be eliminated?
Will it ever be possible to eliminate legacy technology once and for all? Schuerman says he is more optimistic than most.
“The reason I am a little more optimistic is that we have abstracted the business logic from the technology implementation,” he says. That means companies can change their underlying technology without being caught in the “legacy trap”.
“Are we 100% there? No, but that is the vision,” he says.
