Legacy vs. AI in Asset Management: The Real Battle Is Not About Technology

 Asset management firms are talking about AI everywhere - in investment research, client servicing, compliance, operations, and software engineering. Yet the real constraint is not ambition. It is the architecture and its legacy. Many firms are still trying to introduce AI on top of fragmented legacy environments, siloed data, and operating models designed for a pre-AI world.

Ideally, the conversation should not be framed as Legacy vs AI in simplistic terms. Legacy is not merely old technology; it is complexity of technology accumulated over decades, due to pouring budgets into “run the business” activity, leaving only a tiny appetite for a timely enterprise-wide digital transformation which has led to disconnected systems, duplicated processes, partial data lineage. To add to it, there are instances of modernization programmes which has never fully decommissioned the legacy. Many firms continue to carry fragmented technology stacks across asset classes and functions, creating complexity that consumes time, money, and management attention.

And this has now come to bite us, because AI is not just another productivity tool. If used well, it can reshape the economics of the industry. Generative AI, and agentic AI could unlock efficiencies with impacts across distribution, investment processes, compliance and also largely within the Technology units like Software Development. Firms can realize early gains in compliance, risk management, and IT operations, and soon can expand into client-facing and front-office use cases.

But here is the catch: AI does not erase weak foundations. It amplifies them. If data is poor, governance is immature, workflows are broken, or systems do not integrate well, AI will expose inconsistency faster than we imagine. The main barriers to AI value in asset management are cultural resistance, poor data quality, talent gaps, and system integration challenges. Firms increasingly recognize data as critical, yet many still struggle with fragmented systems and outdated processes even as AI adoption moves beyond pilots.

This is why firms should not be approaching AI as a collection of disconnected experiments. They need to treat this as a perfect domain transformation agenda. The outcomes will be stronger if the approach is shifted away from isolated use cases realisations and turn towards moving away from legacy, by embarking an end-to-end redesign of the functions and capabilities to align with the AI world. Long-term advantage of doing so will come from moving beyond superficial adoption and embedding AI into core workflows, governance, and decision-making.

In practical terms, that means three things.

First, modernize the data foundation.

AI cannot operate effectively if firms still rely on inconsistent golden sources, manual reconciliations, and disconnected front-to-back platforms.

Second, redesign workflows - not just tasks.

Real value lies in workflow rewiring and domain-level redesign, not fragmented automation. The biggest returns will not come from deploying copilots in isolation. They will come from reimagining end-to-end processes across research, portfolio construction, onboarding, compliance monitoring, and service operations.

Third, treat governance and workforce readiness as strategic enablers.

Regulatory and compliance complexity will remain if concerns around privacy, accuracy, and external data use remain widespread. Hence the governance structure and workforce readiness needs inclusion at the outset rather than as an afterthought.

So, will AI replace the legacy in asset management? No - not overnight. The debate is not about its replacement. Core platforms still matter for books of records, controls, accounting, transaction integrity, and regulatory confidence. But legacy can no longer be the centre of the strategy. Firms that continue to spend much of their technology energy preserving yesterday’s architecture will struggle to realize tomorrow’s AI value.

The winners will be the firms that do both at once: rationalize the legacy core and build an AI-enabled operating model on top of a modern data and governance foundation. That is the real shift. Not from old technology to the new one, but from fragmented technology estates - to intelligent, integrated, and adaptive enterprises.

Replicated from LinkedIn. Original LinkedIn post - here.

#AssetManagement #AI #GenerativeAI #EnterpriseArchitecture #DigitalTransformation #InvestmentManagement #DataStrategy #OperatingModel #LegacyModernization #FinTech #CIO #CTO #AIAdoption #EnterpriseStrategy

Unlocking AI Value: Why Asset Management Needs an Industry Reference Architecture Now

Further elaborating my previous note, where I touched on why Asset Management industry never embarked into aligning itself to a capability based standard Reference Architecture, here is an attempt to put this into perspective as to why the need is arguably more now than ever.

Asset Management stands at a strategic inflection point. As AI, regulatory scrutiny, and cost pressures converge, the absence of a standard Industry Reference Architecture (IRA) is no longer a technical gap—it’s turning into a business risk and more fragmentation at exactly the time when firms need clarity, consistency, and interoperability.

A well‑defined Industry Reference Architecture (IRA) becomes a force multiplier for AI adoption, regulatory resilience, operational efficiency, and vendor alignment.

Here are some of the strategic advantages - especially relevant when the industry is pivoting heavily into AI, LLMs, and data‑centric operating models

1. AI requires standardisation of data, workflows, and controls to scale

AI does not operate well on fragmented architectures. A capability driven reference architecture provides:

• Common data domains (securities, positions, orders, benchmarks, ESG, alternatives, reference data)
• Standardised process boundaries (investment process, trading, operations, compliance, reporting)
• Clear control points for model governance (model risk, data lineage, human-in-the-loop)
• Consistent integration patterns for embedding AI agents and copilots

The biggest blocker to AI at scale today is architectural inconsistency, not model capability. A standardised reference architecture - directly would be able to remove that blocker.

2. Accelerates interoperability across the vendor ecosystem

Many of the important Asset Management capabilities are very heavily vendor driven. A reference architecture provides:

• Standard sets of integration patterns for all vendors to adhere to
• Standardised capability map to evaluate vendor fit
• Re-usable APIs and canonical data models

This reduces the cost and complexity of vendor replacement or multi-vendor strategies.

3. Creates a shared language between business, technology, and regulators

As AI in financial services becomes more regulated (EU AI Act, SEC guidelines, UK AI White Paper), authorities increasingly expect:

• Explainable process boundaries
• Data lineage
• Governance layers
• Integrated risk and control frameworks

An alignment to a standardised architecture becomes the blueprint to demonstrate compliance.

4. Accelerates AI maturity across the entire value chain

AI + IRA could potentially provide a -

• A unified data fabric
• Clear component definitions (e.g., research, trading, risk, distribution, client solutions)
• Reusable AI patterns (models, embeddings, agents)
• Governance-by-design

This means that instead of “AI in pockets,” the industry moves toward AI-enabled enterprise platforms.

5. Reduces complexity and cost in legacy simplification

A reference architecture -

• Sets the target state to optimize excessively duplicated capabilities and data flows
• Simplifies transformation sequencing by removing inconsistent integration mechanisms
• Allows firms to modernise incrementally without losing coherence
• Enables standardisation of redundant reporting and reconciliation tools

6. Guides AI operating model redesign

Asset Management firms are already exploring options for AI-assisted research, investment idea generation, automated compliance checks, intelligent operational exception handling and more. Industry architecture potentially introduces a baseline to help define the right use cases fit for AI.

Without a defined architecture, AI adoption will remain more ad-hoc, becoming a series of siloed experiments with no enterprise coherence.

7. Strengthens industry collaboration

An IRA becomes a foundation for Industry data standards, Shared solutions (KYC, ESG, market data utilities), benchmarking, and Interoperable best practices of digital ecosystems. This has the potential to help lay a foundation for the Industry Cloud Platforms.

Within the wider Financial Services industry, Banking has standardized operations around BIAN and insurance around ACORD. Asset Management Architecture has the opportunity now for a similar push.

Now is the most strategically aligned moment

For every Asset manager to understand the need to rethink its architecture as:

• Data maturity is becoming a competitive weapon
• Technology costs are rising
• Regulators are expecting more transparency
• Operating models are being rebuilt to be AI-first

A consistent enterprise language in the form of an Industry Architecture is not just a documentation exercise, but a strategic enabler of the next decade of transformation. It is no longer about preserving the past—it’s about shaping an AI-native future. The next decade of asset management will be shaped by those who build AI at scale.

Original LinkedIn post - here  

#ArtificialIntelligence #AssetManagement #DigitalTransformation #EnterpriseArchitecture #AIAdoption #Innovation #Leadership #CTO #CIO #EnterpriseStrategy #IndustryArchitecture #ReferenceArchitecture