The recent $30 billion stock loss IBM faced paints a bigger picture beyond a disastrous afternoon in the wake of Anthropic’s comment. What it also demonstrates is a fundamental misunderstanding that legacy modernisation is simply a technology problem.
For context, AI has been widely treated as a magic bullet for modernisation efforts of legacy systems. However, AI cannot actually bridge the gap between old code and new product architecture. It is not a solution for total transformation.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
When banks focus on swapping old code for newer versions in the bid to modernise their systems, they fall into a translation trap. The reality is that no Tier-1 bank wants to simply translate its proprietary COBOL into proprietary Java. New code will still come with the same maintenance and operational challenges as its 50-year old counterpart. The existing risk and technical debt are also transferred with that code translation.
Ultimately, modernisation failures happen not because of outdated, aging code. The real culprit is a failure to look at the wider blueprint and how emerging AI solutions alongside human expertise play into that. Only from there can banks overcome this hurdle.
Knowledge loss is the big worry
Decades-old code is still in use, and COBOL is the most renowned one. But platforms functioning on old code come with deep business and operational rules that can be impossible to unearth, thanks to missing paper trails and leaking knowledge as data engineers retire.
Lost knowledge is lost money. In fact, because it is treated as an “afterthought,” siloed knowledge is actually costing businesses more than millions of dollars thanks to lost productivity, poor efficiency, undermined collaboration, and operational blocks. Missing knowledge stalls modernisation strategies — there is simply no way forward when banks cannot even understand how their own systems work.
Against this backdrop, any change can multiply problems, including disrupting vital processes such as payments. This is an untenable risk as banks face an exodus of customers who are becoming increasingly exacting in their expectations around digital experiences.
Since banks cannot safely interpret what already exists, legacy systems and their attached weaknesses and issues continue. No matter how sophisticated the technology, without access to the crucial foundational knowledge of existing platforms, all modernisation strategies will be counterproductive.
AI is not just a coding tool
It’s no secret that generative AI has caused big waves in the world of programming and coding, with tools taking on up to 40% of coders’ workloads. Yet AI should not be pigeon-holed for this purpose; by doing so, banks put themselves at risk of leaning on tools for transformation that cannot bridge gaps between legacy and modern architectures.
Instead, AI integration should hinge on complementing human expertise and optimising efforts. One of the best avenues for that is by leveraging AI solutions for systems interpretation, switching the tools’ role from ‘coder’ to ‘systems historian.’
Nobody writes a core system from scratch, and decoding an existing one takes a huge amount of time and effort. Sifting through decades’ worth of data and rules buried deep within an older framework is, at best, an uphill battle and, at worst, impossible. AI, though, can be used to analyse legacy codebases to uncover hidden dependencies and business rules. It can help teams uncover the intent behind systems, not just their structure.
AI as a coding assistant has led to productivity gains, but these are a far cry from its real potential. Integrating these tools to delve into the history of a system enhances the foundational understanding needed for successful modernisation strategies.
Banks can successfully map and discover systems, more effectively spread out expert resources, which are becoming few and far between, and ensure better-informed decision-making ahead of migrating or replacing a system. Moreover, this ensures success is sustained and scaled, not short-lived.
Orchestration outweighs ownership
As it stands, banks are losing interest in ownership. Legacy systems come with strings attached that are proving to be a strain on effort and manpower, and in fact, banks want to offload the burden of maintenance. As long as the gaps between old systems and new architectures exist, the core strategic question remains: why would a bank want to own code? Most banks don’t want to be software houses for proprietary code they can’t afford to keep.
Because of this wider trend of outsourcing expertise and integrating external platforms, the immediate aim is not actually to rebuild code but to orchestrate systems. Orchestration is at the heart of successful digital transformation, where systems scale in tandem with organisational needs. It is a matter of what to keep, what to replace, what to discard, and what to orchestrate.
AI has the important role of accelerating transformation efforts by helping organisations understand legacy systems to achieve orchestration. These tools should be used for building the roadmap of a bank’s digital modernisation by making complex systems easier to understand, no matter how old they are. Right now, AI’s most important role is helping banks understand what they already have before deciding on what comes next.
Claudio González, Global CTO & EVP at intive
