On May 12th, we marked a significant milestone for Zitec by hosting our first-ever international roundtable in Amsterdam, organized in collaboration with the Holland Fintech Association. The exclusive, intimate gathering brought together financial services leaders for an in-depth session titled "How to Identify & Govern AI Projects in Financial Services".
Rather than repeating the usual industry hype about how AI will magically reinvent financial services overnight, this focused environment allowed us to have a candid, high-impact conversation grounded in reality. We looked closely at what happens when the initial excitement of a proof-of-concept (PoC) meets the reality of enterprise infrastructure.
Generative AI is establishing itself as a core pillar of digital transformation within financial services, driving documented returns on investment above 20%. The key to achieving these results lies in moving away from generic use cases and focusing heavily on specific, manually-intensive processes.
During the roundtable, we mapped out distinct scenarios where financial institutions are seeing measurable operational improvements:
Automating data extraction from complex financial statements, tax returns, and legal contracts can allow teams to reduce manual entry errors, speed up requests processing time (such as loan requests) and strengthen compliance. This can generate huge increases in lead throughput capacity. For one of our international clients, we experienced a 3x increase in lead-processing capacity.
Utilizing generative AI to analyze customer intent and route incoming requests can lower triaging costs by up to 75%, while increasing efficiency and consistency. This accelerates response times and frees human agents to focus on high-value interactions, reducing agent burnout.
By unifying data sources, such as regulations, best practices and historical customer interactions, institutions can help employees onboard easier, all while transferring knowledge in an efficient way. This eliminates information silos, accelerates decision-making, leads to improved customer experience and better compliance with existing processes, which in turn can lower search-related costs by more than 50%, while maintaining answer accuracy.
By deploying generative AI systems that analyze historical company best practices, previously successful pitches, or customer data, organizations can help sales teams mix and match all the successful sales tactics, in order to improve conversion rates and provide personalized sales experiences. This targeted approach helped one of our clients increase conversion rates with 15% and to deepen client relationships.
Drafting and formatting complex regulatory filings across multiple jurisdictions can take up a massive amount of senior team hours. Deploying agentic AI systems to monitor legal updates, generate standardized documentation and identify regulatory gaps, preventing non-compliance issues and potential fines before they occur, can reduce creation and review times by 30%.
When it comes to software development, there is currently a visible gap between what organizations expect AI to deliver, orders of magnitude improvements in time and cost, and the actual marginal gains seen on the ground.
Data from the recent DX Report, which evaluated over 400 companies employing more than 100 engineers, highlights that most enterprise organizations are experiencing an overall AI productivity benefit of somewhere between 5% and 15%.
Why do we see such a massive discrepancy between enterprise results and the viral stories of solo founders building massive apps over a weekend? It comes down to the shape of the work. A solo founder operates with zero coordination overhead. They have no architecture boards, no compliance checks, and no legacy code to maintain. Not to mention the survivorship bias in such stories.
Enterprise engineers, by contrast, spend their days maintaining complex systems. They must integrate new tools with dozens of existing services and support legacy API contracts. When you deploy AI onto a small, fresh codebase, the results feel transformative. When you deploy it onto a 2-million-line legacy monolith with 14 years of accumulated context, you get a realistic 5% to 15% marginal gain. AI is an amplifier, but it cannot automate away systemic enterprise overhead.
To move beyond the limitations of isolated prototypes, financial institutions must adopt a structured framework that requires a dual-track alignment: building a resilient, compliant technology architecture while simultaneously reshaping operating workflows to support automated, agentic systems.
A financial institution cannot build a reliable AI application on top of a fragmented, siloed data landscape. True readiness requires a unified, trusted data estate combined with secure, compliant connectivity to core enterprise systems. Establishing an explicit AI technical strategy and an optimized cloud landing zone ensures that applications remain stable, while developing reliable technical partnerships, including collaborations with EU-based technology providers, guarantees regional data sovereignty and regulatory compliance. .
Organizational readiness moves the conversation from infrastructure to execution. Successfully operationalizing AI demands active executive sponsorship and the cultivation of an embedded, in-house AI competency, which can be strategically complemented by specialized external technology partners. Organizations must maintain a cultural willingness to allow structured experimentation, meaning they must actively adapt traditional company processes to accommodate modern, agentic workflows where human oversight and automated intelligence closely collaborate.
Traditional procurement methods must be replaced by rigorous use-case prioritization. To build a balanced portfolio, institutions should utilize a selection process driven by cross-departmental brainstorming, scenario curation, and strict business impact alignment. This stage requires meticulous outcome measurement and ROI planning that accounts for tokenomics, a historical lack of baseline data, and the natural variance in returns across different degrees of process automation.
Because AI project returns carry high variance, fixed budgeting models often constrain performance and stifle innovation. Financial institutions excel when they approach portfolio management through dynamic funding, intentionally avoiding rigid annual allocations that do not account for technical discoveries. Instead, budgets must be managed dynamically, unlocking capital through performance-linked tranches that adjust automatically as a specific use case proves its technical feasibility and delivers tangible business value.
At Zitec, we have spent the past 23 years building digital solutions for industries where software simply has to work without compromise, delivering over 1,100 projects across financial services, logistics, healthcare and retail. We know from experience that successful digital transformation is never about adopting technology because it is fashionable. It is about mapping the actual problem and designing processes that fix it, operating reliably at scale.
If you are currently evaluating your portfolio of AI initiatives and want to separate the practical use cases from the noise, reach out directly to our financial services team and let’s have a chat.