Organisations in the banking sector are increasingly moving away from fragmented legacy systems towards unified, scalable AI platforms that can deliver measurable business value. Mohamed Galal, Head of AI and Data Management at National Bank of Egypt, tells us how adopting a centralised platform has improved governance, accelerated deployment and enabled more effective collaboration between business and technical teams.
What were the biggest challenges you faced in maintaining legacy models, and how did the transition to a unified platform like Dataiku help address those issues?
One of the biggest challenges we faced with legacy environments was fragmentation. Models were developed across multiple tools and teams, leading to inconsistent governance, duplicated efforts and limited scalability. This made it difficult to maintain models efficiently, ensure compliance or adapt quickly to evolving business needs – particularly in a highly regulated banking environment.
Another key challenge was the disconnect between business and technical teams. While strong models were being developed, translating them into sustainable, production-ready solutions was often slow and complex. This limited our ability to scale AI beyond isolated use cases.
Transitioning to a unified platform like Dataiku fundamentally changed this dynamic. It provided a centralised environment that integrates data, development, deployment and governance in one place. This allowed us to standardise processes, ensure consistency and significantly improve model lifecycle management.
Most importantly, it enabled us to shift from fragmented experimentation to a structured, end-to-end AI operating model – where solutions are built, governed and scaled with business impact in mind.
How has the adoption of Dataiku improved collaboration between technical and business teams across the bank?
Collaboration is one of the most transformative impacts we have seen. Traditionally, AI initiatives were largely driven by technical teams, with limited business involvement beyond defining initial requirements. This often created gaps between model outputs and real business expectations.
With Dataiku, we established a shared platform where both technical and non-technical stakeholders can actively contribute throughout the AI lifecycle. This aligns perfectly with our internal model of ‘AI ambassadors’ – business representatives embedded within each domain who work closely with data teams.
Through this approach, business teams are no longer passive stakeholders; they participate in defining use cases, validating models and ensuring outputs are aligned with real operational needs. At the same time, technical teams benefit from clearer requirements and faster feedback loops.
The result is a true Human-AI collaboration model, where AI becomes a shared organisational capability rather than a centralised function. This has significantly improved adoption, trust, and the overall effectiveness of our AI initiatives across the bank.
Can you quantify the impact the new platform has had on development speed and the time required to operationalise new use cases?
While exact figures may vary by use case, the impact on speed and efficiency has been substantial. We have significantly reduced the time required to move from idea to production by streamlining development, testing and deployment processes within a single platform.
Previously, operationalising a model required co-ordination across multiple environments, manual integrations and extended validation cycles. Today, with a unified platform, these steps are integrated and automated, enabling faster iteration and quicker deployment.
In practical terms, this has allowed us to accelerate our AI time-to-market and scale multiple use cases in parallel rather than sequentially. More importantly, it has enabled us to focus less on technical overhead and more on delivering measurable business value.
This acceleration is critical in banking, where responsiveness to customer needs, market changes and regulatory requirements can directly impact competitiveness. Ultimately, the platform has transformed AI from a slow, project-based activity into a continuous, scalable capability.
How important was governance and MLOps capability in your decision to modernise your data and AI environment?
Governance and MLOps were not just important – they were fundamental to our decision.
In the banking sector, AI must operate within strict regulatory, security and ethical boundaries. Without strong governance, scaling AI can introduce significant risks, including lack of transparency, bias and compliance challenges.
From the beginning, we recognised that sustainable AI requires governance by design, not as an afterthought. This includes model traceability, explainability, bias monitoring, version control and clear auditability across the entire lifecycle.
MLOps capabilities were equally critical. To move from experimentation to production at scale, we needed robust processes for deployment, monitoring, retraining and performance management.
Dataiku provided both elements in an integrated manner, allowing us to embed governance and operational discipline into every stage of development. This gave us the confidence to scale AI responsibly while maintaining trust with regulators, stakeholders and customers.
Ultimately, governance is what transforms AI from innovation into a reliable, enterprise-grade capability.
With the foundation now in place, how are you planning to scale into more advanced use cases such as AutoML and Generative AI?
With a strong foundation in place, we are now strategically expanding into more advanced AI capabilities, including AutoML and Generative AI.
Our approach is structured and deliberate. First, we ensure that these technologies are aligned with clear business use cases – whether enhancing customer experience, improving internal efficiency, or supporting decision-making. We do not adopt AI for experimentation alone; every initiative must deliver measurable value.
Second, we leverage the platform’s capabilities to democratise AI further. AutoML enables faster model development and empowers a broader range of users, while maintaining governance and control. This allows us to scale AI adoption without compromising quality.
On the Generative AI side, we have already started implementing use cases and are exploring more advanced concepts, including Agentic AI. These capabilities are being integrated within our governance framework to ensure responsible and secure deployment.
Our focus is not only on innovation, but on sustainable, enterprise-wide impact – ensuring that advanced AI becomes a core driver of business value.
What lessons would you share with other financial institutions looking to move away from legacy systems towards a more unified and scalable AI platform?
The most important lesson is that AI transformation is not just about technology – it is about strategy, people and operating models.
First, align AI with business objectives. AI should not exist in isolation; it must be directly linked to measurable outcomes and integrated within the broader Digital Transformation strategy.
Second, move beyond isolated use cases and think in terms of end-to-end processes. At NBE, shifting to a full customer journey perspective – acquisition, development and retention – was a key enabler of scale and impact.
Third, invest in people and collaboration. Building AI ambassadors within business teams was critical to bridging the gap between technical and operational domains.
Finally, choose the right platform strategically, not just technically. A unified platform with strong governance, scalability and collaboration capabilities is essential to move from experimentation to enterprise-wide adoption.
Organisations that successfully combine these elements will not only adopt AI – they will position themselves to lead in the future of banking.
