AI & ML
built in the 1970s” – a vivid illustration of the technological gap many financial institutions face.
“ These systems, while robust, were never designed for the realtime, API-driven world of modern finance,” Gunnar explains. Rather than completely replacing these systems, banks are deploying“ API gateways and microservices as the digital‘ translation layers’ between old and new,” enabling gradual modernisation while maintaining operational continuity.
Gunnar elaborates on this approach:“ By encapsulating core banking functions into modular, scalable microservices, institutions can gradually transition towards a cloud-native architecture, enabling real-time processing, enhanced security and seamless third-party integrations.”
The most forward-thinking banks are adopting what Gunnar calls“ progressive modernisation, where legacy systems don’ t get ripped out overnight but are gradually phased out while API-first components take over key functionalities.”
The evolution toward AI-powered APIs Looking ahead, Gunnar envisions APIs evolving beyond simple data transmission to provide AI-driven insights.“ Imagine an API that doesn’ t just retrieve account balances but analyses spending behaviour, predicts cash flow issues and suggests personalised investment opportunities in real time.”“ With AI models embedded within APIs and with more data available at the time of transaction, financial institutions are more likely to predict fraud before it happens, automate risk management and deliver personalised financial insights to customers,” Gunnar predicts. This evolution represents a fundamental shift in how financial services operate.“ The move from basic APIs to smart, AI-powered ones means banks won’ t just respond to customer needs- they’ ll predict them in advance,” Gunnar says. The institutions that leverage AI-driven APIs will redefine customer relationships, moving from transactional interactions to deeply embedded financial advisory.
As these technologies advance, privacy concerns become increasingly important, particularly in light of regulations like GDPR and open banking initiatives. Gunnar advocates for“ privacypreserving computation techniques” such as tokenization, zero-knowledge proofs and homomorphic encryption to protect sensitive information.
“ Adopting privacy-preserving computation techniques ensures that sensitive information can be computed on without ever being exposed,” explains Gunnar. Meanwhile, consent mechanisms embedded into APIs ensure customers maintain control over their data.
“ Fine-grained consent mechanisms must be embedded into APIs, ensuring that customers have full control over who accesses their data, for what purpose and for how long,” Gunnar concludes, highlighting the critical balance between innovation and privacy that will shape the future of predictive analytics in finance.
118 April 2025