They say change is constant and this statement holds true for new-age technologies. Every day there is a new progress in software technology, turning to a new page in modernization, and that relentless progress, sadly, comes with a lurking threat, i.e., complexity.
It is a known fact that legacy systems, once marvels of ingenuity, can transform into a tangled mess of code, defying comprehension and may even hinder further innovation. This is the tightrope walk that most CTOs and CIOs face - navigating the terrain of software modernisation, where a single misstep can cripple operations or cause monetary loss to the organisation.
The path to the cloud beckons, promising agility and scalability, but the journey is undoubtedly fraught with the perils of complexity in terms of data privacy. To embark on this modernization journey, organizations must confront the software complexity head-on, wielding the combined power of human expertise and Artificial Intelligence (AI).
While much of the nuances of dealing with software complexity is not known beyond the walls of software giants' establishments, BW BusinessWorld took the initiative to organize a Round Table Conference to demystify and bring to fore the various complexity issues that organizations face at their end.
Supported by CAST, the Paris-headquartered software intelligence company, the round table conference opened up critical viewpoints to tackling software complexity in the face of modernization. Fifteen C-level executives participated in this round table discussion that was ably moderated by Vincent Delaroche Founder & CEO, CAST and Hoshie Ghaswalla, CEO & Managing Editor, BW Businessworld.
The in-person Round Table, held on 24th April, at Bengaluru, offered a great forum for all participants to brainstorm and unlock varied insights on the topic of Tackling Complexity: The convergence of AI and software intelligence on modernizing and transforming systems
Synopsis from the round table discussion:
Taking measured approaches in software modernization
Organisations, today, operate in a landscape that's more complex than ever. Especially, for an organization in the financial services sector, software governance plays a very important role and modernizing software in this business would be subjected to scrutiny of all kinds. Data privacy takes superior importance and while moving from legacy systems, it is essential to bear in mind that no compromise can be done with handling data.
Throwing an umbrella question open to the forum, Hoshie Ghaswalla, initiated an avalanche of purviews. He asked, "How does your organization ensure the preservation of software knowledge, including utilizing AI-driven code analysis, during modernization efforts and transitions to the cloud?"
This question was asked in the context of CIOs, today, grappling with the intricate software backbone vital to their operations, the complexity surpassing human capacity to make the invisible visible, necessitating AI code analysis for software knowledge retention during modernization and cloud migration.
Responding precisely to the question, Chakra Mantena, Managing Director and Head of Technology Global Centres, Morgan Stanley, was the first to answer and got the session rolling. He said, "The financial services industry landscape is constantly evolving, with new regulations, technologies, and customer demands. This makes it essential for organizations to continuously innovate and evolve their technology platforms, and maintain strong governance to ensure that software development and deployment are aligned with the organization's overall business goals, regulatory requirements, and security standards.
Commenting further, Mantena, said, "In the process of modernization, writing code is important, but comprehensively testing code is equally important. Rigorous testing builds confidence in the modernized system. It provides evidence that the new system meets business requirements and functions reliably before deployment. This is crucial for all the stakeholders in the ecosystem."
In agreement with Mantena's views on software complexity, Sunil Gopinath, CEO of Rakuten India, stated, "For companies like Rakuten, we have different businesses that have started at various stages-whether 20 years ago, 10 years ago, or 5 years ago. So, there is a generation of complexity built into the system. This means the tech stack and data needed for every business is different, giving a new twist to complexity. Navigating this landscape involves precise decisions about when and how much of our systems to migrate or whether to undertake a complete re-architecture."
Gopinath highlights a particular challenge in finding talent adept in legacy programming languages like C++. "The rarity of expertise in certain older technologies is where we leverage Generative AI to refactor legacy code, integrating modern advancements back into our systems seamlessly," he states.
Yoginder Grewal, CIO, Hindustan Coca-Cola Beverages, giving another pertinent perspective to the question, said that while his company gives importance to deploying AI at different stages, especially in data analytics, he believes that investing in the right people is key to successful outcome of AI-led modernization. Grewal believes that partners involved in development need to have right skillsets/experience else the partner would end up learning at the client's cost.
Can AI fulfil business agility requirements?
AI can be a powerful tool to support business agility, but it is not a silver bullet. AI excels at analyzing massive amounts of data to identify trends and patterns but the AI systems still require human guidance and manual oversight to ensure they are functioning correctly and ethically.
Ganesan Ramani, Senior Vice President - Head of Technology GRC & TMO, Mashreq, said, “Financial industry is the most regulated industry. Any new technology adoption would mean we need to evaluate thoroughly on the regulatory compliance requirements and the set of rules that govern its operations. And the use cases of AI in the financial industry are plenty. Chatbots support in digital banking experience and these bots are now powered by AI to give a modern, customized, and personal banking experience to the customers.”
"Financial services firms are burdened by legacy technologies, disparate data sources and manual processes. To add to that, mergers and acquisitions that are quite common in this industry, further complicate the underlying technology architecture. AI and other emerging technologies are promising in terms of improving developer productivity, generating new insights, eliminating manual processes, improving data validation and error detection. The ability to parse unstructured data and documents open up avenues to perform valuation or cash flow analysis on alternative asset classes effectively, flag suspicious KYC/AML transactions more accurately. With GenAI chatbots a trader/ analyst could use natural language to query about portfolios, counterparty risk exposure etc. While many financial services firms are on their journey to adopt these new technologies, in reality they will continue to exist in a symbiotic relationship with the legacy technologies" opined Neetha Raja, VP, State Street.
The question of AI driving business agility remains to be addressed but is a long-drawn process. It is for certain that AI's role in driving business agility is a complex and evolving concept. Though AI holds immense potential, it is not a magic solution and realizing its full impact takes time and effort.
Voicing his opinion from a textile industry's context on bringing down software complexity and the deployment of AI to modernize software architecture, Satish Panchapakesan, Sr. Vice President & CIO, Arvind Fashions Limited, said, "The key takeaway for businesses of our kind from any AI/ML tool is to look at how it will enhance business agility. We are concerned about how these tools can take the data quotient to the next level. There is a lot of data from interfaces, applications, etc. falling into a large sink or a lake. How is right meta data governance built around this data lake is a problem to be addressed by AI. The other looming question is how good are we in turning interfaces consistent to existing software architectures using AI. The AI system must be able to give granular data analysis to take appropriate business decisions at the right time is also key. The AI tool must also have the ability to bring about an agile way of replicating the interface in a consistent way in attaining revenue generation and being cost efficient at the same time. The third most important expectation from an AI system is the differentiation that it can bring about in the overall business."
Stepping in at this juncture, Vincent Delaroche, Founder and CEO, CAST, brought to fore the advantages of deploying software intelligence systems. He explained, "AI models require large amounts of labeled training data to be effective. Analyzing millions of lines of code, especially proprietary code, can be challenging due to limited access to training data. Understanding the rationale behind AI-generated insights, particularly in complex code analysis, can be difficult. This can hinder trust and adoption. Software intelligence tools here play an important part in the software lifecycle continuum. These tools can offer a more comprehensive view of the codebase, analyzing factors like code complexity, maintainability, and potential security risks."
Modernizing - Not an event but a journey
Technology is constantly changing. New tools, frameworks, and best software developing practices emerge all the time. And in this milieu, modernization ensures the software stays relevant and leverages these advancements. "But it is easier said than done. Rewriting millions of legacy codes is not an event it is a journey," vehemently expressed, Mayur Purandar, Vice President, Enterprise Architecture, Lowe's India.
Purandar further said, "Big bang software transformation or migrations is rarely successful. The biggest challenge in legacy software modernization comes while iterating the modernization journey, where the legacy software needs to run parallel to the modern software. We need to peel off one layer to add a new layer to the upgrading software ensuring that the code functions are not altered. What are my 20 interim steps that I must pick up from the layers of left to plug into the right not compromising on functionalities puzzles us always."
Purandar observes that the entire paradigm of software working has changed from mainframes to now. Mainframes was time bound whereas currently the software systems are real time or near to real time, making workings of the software very different from before.
Offering more views from his experiences to the context in play, Samir Gulve, VP Engineering & Managing Director - India, EFI, said, maintaining software is a necessary evil. “The first question when it comes to modernization is when to do it and to quantify the transformation needs. Answers to these will give cost-effective solutions. AI definitely has the ability to solve non-deterministic problems. So, to tap its potential can go a long way in software development lifecycle. And in the process if one can determine if the decision-making abilities of the software can be improved then it can be logically inferred that transforming or modernizing is inevitable. The most important requirement is the availability of quality and relevant data to allow effective transformation.”
Prabhas Abhayakumar, VP at Societe Generale Global Solutions Centre feels that the industry has seen a steady progress in automation by leveraging technology such as RPA, API and AI/ML tools. Tools such as Automation Anywhere is being adopted to create BOTs that automate many manual tasks on the operations side, which has helped in reducing significant manual effort and operation risk as well. The focus is also to reduce and rationalise application landscape by building platforms which will expose functionality as services. These platforms also leverage Machine Learning capabilities that harness the vast amount of data that gets generated across various business operations. The focus is also to improve on the quality of data which in turn helps in accurate decision making and proper reporting. What is next in the transformation journey is the big question. To this, Prabhas Abhayakumar strongly feels the focus is improving the productivity of the IT teams by leveraging AI tools such as ChatGPT and Microsoft Co-pilot. The success of this adoption will depend on business use case that are able to demonstrate significant savings in cost and reduction of risk, at the same time being able to manage data accuracy and privacy concerns.
Modernising with a judicious approach
Sony Pictures Networks India has strategically approached its digital transformation, prioritizing business value in every decision. Raj Mohan Srinivasan, explains, "We focus on meaningful changes, opting for low-code/no-code solutions and integrating AI and ML to enhance efficiency and reduce costs". He notes significant savings from automating processes, moving to a cloud-based platform, and shutting down physical data centres. Srinivasan also highlights the controlled use of Gen AI and other tools for managing deepfakes, ensuring content integrity. This approach exemplifies the company's commitment to practical and impactful technological advancements.
Well, Sony Pictures moved its platform to cloud. But, is moving to cloud a good option for all? Amit Khanna, Associate VP Engineering, Amadeus Labs, said, asserts, “The decision to migrate to the cloud hinges on organizational size and strategic objectives. Amadeus made a deliberate choice to embrace the cloud, architecting a scalable platform. This transition is proving to be a game-changer.”
Logic says, that from a cost perspective, transformation to business process needs to be done only if there is a significant RoI generation envisaged. The metrics involved in this transformed have to be clearly spelt before moving ahead. Sharing some very specific numbers Ravindra H S Vice President - Engineering & India Site Leader, Pluralsight India, said, "From what Pluralsight has gathered, only 6% of RoI in developer productivity are yielded. Unless this transformation gives 13-14% RoI in developer productivity, there is no value-add that one can derive."
Sharing a sunny side to the use of AI technologies, Raj Lakkundi, VP of Software Engineering, Epsilon, said," As AI continues to advance, companies that embrace these technologies will have a competitive edge in the market. Tools like GenAI significantly upskill developers' ability to write better code. We have also observed an increase in the creativity of engineers, saving us time and money while continuing to push the boundaries of technological advancements. Gen AI and software intelligence go hand-in-hand "
Bringing in a new perspective to the discussion and sharing about the software related complexity in the healthcare industry, Dileep Mangsuli, Executive Director, Siemens Healthineers Development Centre, said, “The healthcare industry is the most regulated especially with respect to data privacy. Although we use AI in our imaging and diagnostic products, they only assist in data analytics and research. The future of AI intervention in healthcare will be in predictive and preventive healthcare. AI and Gen AI will also help in post-sale customer servicing of healthcare equipment through chat bots and other product maintenance related warning signs. AI-led digital twinning of organs will be one area where there will be a lot of excitement in the coming years.”
New incumbents/FinTechs in the market tend not to have the baggage of legacy systems. Organizations of a medium scale that are in the initial stages of their tech adoption journey have less to worry about software upgrades. Shashank Agrawal, Chief Technology Officer, HDFC Credila Financial Services Limited, believes in adopting a focused and incremental approach when dealing with software complexity to drive long-term transformation while ensuring optimal ROI on tech spending. "We have identified pockets within platforms that can be augmented by leveraging Gen AI tools for workflow optimization, loan document processing automation, as well as predictive analytics, to delight customers, increase efficiencies, and support sustainable growth," said Agrawal.
Mantra of pragmatism in tackling complexity
Tackling software complexity, we know, is not easy. This is where software intelligence systems play a valuable role in modernizing software by providing a deeper understanding of existing codebases. In a well-explained verdict to the discussion CAST's Vincent Delaroche, said, "Software intelligence Platforms are the need-of-the-hour. These Platforms can analyze millions of lines of code, complexity metrics, and potential security vulnerabilities. This comprehensive picture helps an organization prioritize modernization efforts. Technical Debt Assessment can be quantified, outlining the hidden costs of maintaining outdated or poorly written code. Also, these Platforms can assist in migrating code to newer frameworks or platforms by identifying potential compatibility issues and suggesting solutions. I can say that adopting these intelligence platforms is the most pragmatic way to derive data-driven insights in making strategic decisions throughout the modernization journey. By combining these platforms with human expertise, the software development cycle can be maintained to yield sustainable cost-effective business practises.