Capital-markets firms are modernizing the trade lifecycle under pressure from every direction: growing data volumes, higher expectations for real-time insight, AI initiatives moving toward production, and shorter settlement cycles. At this point, nobody's debating whether to use AI. The real question is whether research, trading, risk, ops, and compliance are all working off the same governed data — or just telling themselves they are.
As firms move from experimentation to production, a pattern is emerging: the durable advantage is not a model in isolation. It is the ability to make proprietary data, like orders, executions, positions, research, risk, client, and operational data more discoverable, reliable, and governed across the lifecycle.
I spoke with Andrea DeSosa, Global Head of Capital Markets GTM at Databricks, about the operational pressures reshaping pre-trade, execution, post-trade, and surveillance workflows, to get insight on practical ways to prioritize modernization. This conversation has been edited for clarity and length and draws on themes explored in the ebook Modernizing the Trade Lifecycle in Capital Markets.
Why trade-lifecycle pressure is cumulative in 2026
Kim Hatton: What is forcing trading and research leaders to revisit the trade lifecycle now?
Andrea DeSosa: Firms are managing more alternative data, more AI experimentation, shorter operational timelines, and closer scrutiny of how decisions are made and controlled. At the same time, research, trading, risk, operations, and surveillance often rely on separate systems and inconsistent data views.
That fragmentation slows decision-making, complicates execution analysis, limits risk visibility, and makes it harder to reconstruct what happened when a regulator, client, or internal control function asks.
Kim Hatton: Where does data fragmentation show up in the trade lifecycle?
Andrea DeSosa: In pre-trade, teams can spend substantial time preparing and reconciling market, fundamental, ESG, and alternative data before they can test an investment hypothesis or run a backtest.
During execution, desks need timely views of execution quality, liquidity, order flow, and slippage by venue, strategy, and client. But the view on a trader’s desktop may not reconcile with the data available to risk, compliance, or operations.
Post-trade, many firms still depend on overnight risk processes, spreadsheet-based reconciliations, and reporting designed for a longer settlement cycle. Those practices become harder to sustain as settlement timelines compress and expectations for operational resilience and transparency rise.
Where AI fits in trade-lifecycle modernization
Kim Hatton: Where does AI fit into trade-lifecycle modernization?
Andrea DeSosa: The firms seeing repeatable value from AI are not necessarily the firms running the largest number of pilots. They are the firms that can securely connect and govern proprietary context—order flow, internal research, positions, risk outputs, client information, and operational data—across the lifecycle.
Without that foundation, AI can create another disconnected workflow. With it, firms can build tools that help employees investigate exceptions, analyze execution quality, synthesize research, or surface potential surveillance issues while maintaining appropriate access controls and traceability.
The strategic question shifts from “Which model should we use?” to “How do we connect our data, analytics, and controls into production workflows?”
Three signals it's time to modernize the trade lifecycle
Kim Hatton: What signals mean it's time to modernize trading architecture?
Andrea DeSosa: Three come up repeatedly.
- Shortened settlement cycles. Increase the need for timely, reliable views of trade status, funding, collateral, exceptions, and operational risk. In Europe, the move to T+1 settlement is scheduled for 11 October 2027, making readiness a near-term planning priority for firms active in EU and EEA markets.
- Governance and auditability requirements. Firms need to demonstrate how automated and AI-enabled workflows are governed, tested, monitored, and investigated end to end. That requirement spans data quality, entitlements, lineage, model and agent evaluation, decision records, and the ability to investigate exceptions.
- Avoiding shadow AI tools. Teams want to put AI into research, execution analytics, operations, and surveillance workflows without creating shadow data sets, unmanaged tools, or inconsistent controls.
These pressures lead to one practical test: when rates move, spreads widen, liquidity deteriorates, or an operational incident occurs, how quickly can the firm see its positions, understand exposures, identify affected workflows, and produce an auditable account of the data and decisions involved?
How to start trade-lifecycle modernization: decisions before migration
Kim Hatton: Where should firms begin trade-lifecycle modernization?
Andrea DeSosa: Start with decisions leaders already need to make:
- Where is execution cost diverging from expectations by venue, strategy, and client?
- How would a rate or volatility shock affect risk and liquidity by desk, region, or portfolio?
- Which business lines, asset classes, or counterparties are associated with the highest exception, break, or settlement-fail rates?
Once the priority questions are clear, the required data domains become clearer too.
For execution-quality analysis, that may mean market and reference data alongside orders, executions, venue data, and benchmarks. For post-trade resilience, it may mean trades, allocations, confirmations, settlement status, collateral, funding, and operational exceptions.
The goal is not to modernize every system at once. It is to establish a governed, reusable data foundation for a small number of high-value workflows, prove measurable outcomes, and expand from there.
How Databricks governs AI across the trade lifecycle
Kim Hatton: What does the Databricks Data + AI Platform enable in this approach?
Andrea DeSosa: The platform supports that progression by bringing real-time and historical data, analytics, and AI together on a common foundation.
Unity Catalog provides centralized governance, access control, discovery, and lineage across data and AI assets. Agent Bricks can help teams build, deploy, and govern domain-specific agents, while AI/BI enables business users to explore governed data through dashboards, visualizations, and natural-language experiences.
The value is not limited to one workflow. It is the ability for research, trading, risk, operations, and compliance to work from consistent, governed data while preserving the controls and context each function needs.
The models will evolve. The business questions, proprietary data, and requirements for trusted controls will remain. Firms that treat trade-lifecycle modernization as a governed data and operating-model initiative—not a collection of isolated AI features—are better positioned to improve decision speed, operational resilience, and the ability to scale new use cases over time.
Learn more
Download our new Ebook Modernizing the Trade Lifecycle in Capital Markets for practical pre-trade, execution, post-trade, and surveillance use cases, and a framework for prioritizing governed data and AI initiatives across the lifecycle.
Instagram Engineering
Shopify Engineering