By Mick Hittesdorf, OneTick Senior Cloud Architect
I recently had the pleasure of participating in the "Rewiring the Front Office, Cloud Data and AI in Action" webinar alongside Peter Ottomanelli of American Century Investments and Vishal Gupta of Mizuho Americas. We dug into the real-world shifts happening on both the buy-side and sell-side as firms transition away from legacy platforms toward cloud-native, API-driven, and AI-enabled environments.
The discussion confirmed that while enthusiasm for AI and the cloud is high, the path to modernization is complex. Here are my top five takeaways from our conversation:
The industry has crossed the inflection point; the question is no longer if we should move to the cloud, but how to do it safely and efficiently. However, many firms are stuck in an "execution gap". Because of the "data gravity" pulling on decades of legacy infrastructure, large-scale migrations don't happen overnight.
Consequently, hybrid and multi-cloud architectures are the default reality for the foreseeable future. Firms are finding success by intelligently decoupling monolithic applications into composable APIs and migrating those to the cloud, rather than attempting massive, risky "big bang" migrations.
It is a universal truth in data science: garbage in, garbage out. However, as Peter rightly pointed out, AI exponentially amplifies this issue. AI dramatically compresses the distance between raw data and execution. If you feed an AI agent bad data, it will fail at machine speed, and by the time you realize something is wrong, it may be too late.
This is why establishing verifiable data trust is paramount. At OneTick, we believe data quality scoring is essential. We are actively working on tools that will surface data quality scores as metadata, allowing both human consumers and AI agents to objectively assess whether the data they are relying on is trustworthy before making a decision.
Python has undisputedly become the language of choice for front-office experimentation. From traders to middle-office personnel, it seems everyone is running Jupyter Notebooks.
While this drives incredible collaboration and fast innovation, it also introduces significant risks if left ungoverned. The proliferation of bespoke Python versions and unsupported third-party libraries can quickly create an unmanageable "shadow IT" ecosystem. Python will remain the language of innovation, but firms must wrap it in rigorous engineering discipline to ensure business continuity.
When we talk about front-office modernization, we naturally focus on speed and scale. But Vishal raised a critical point: is our control architecture evolving as fast as our analytics architecture?.
While some argue that the "human-in-the-loop" concept is becoming outdated as we move toward agentic AI, true accountability still demands human oversight. You cannot hold a machine legally or financially accountable. As firms deploy autonomous workflows, there must still be a human operator with a proverbial "big red button" ready to pause trading or stop an LLM when things go off the rails.
If I had to predict the next three to five years, I see the center of gravity in the cloud shifting. Historically, the primary draw of the cloud was elastic compute—how we pay for it and procure it on demand.
While compute elasticity remains important, the real focus moving forward will be firmly on the data itself. This is why investing in open data formats, like Parquet and Iceberg, is so critical. Open formats provide the necessary interoperability across diverse compute platforms, prevent vendor lock-in, and ultimately future-proof your data strategy, enabling the seamless access to clean data that modern front offices require.
If your firm is looking to overcome legacy data gravity and build a modern, high-performance tick data architecture in the cloud, let's talk. Contact the OneTick team today to learn how our solutions can accelerate your front-office modernization.
Best wishes,