In 2019, Ben Robinson, a leading fintech strategist set out an argument that banking was at a breaking point. Legacy systems of record, built for a branch-based world, were creaking under the weight of modern customer expectations, fragmented channels, and the growing demand for embedded finance. His prescription: a new architectural layer, systems of intelligence, designed to mediate between systems of record and systems of interaction.
Fast forward to 2025, and in his recent article “Digital era banking systems revisited”, Ben reflects on the theory with the central question: has the world moved towards systems of intelligence, and what role does Generative AI play in accelerating (or complicating) this shift?
His conclusion is clear: systems of intelligence are not only still relevant, but more essential than ever. And from additiv’s vantage point, working with leading banks, insurers, and asset managers globally, we couldn’t agree more. In fact, we see orchestration platforms as the decisive factor in whether incumbents can keep pace with digital-native competitors and truly capitalize on AI’s potential.
Here, we unpack that thesis, reflecting on how it aligns with what we see in the market, and where systems of intelligence can deliver the greatest impact.
The case for systems of intelligence
As we know all too well, and highligted in the article, most financial institutions still run on decades-old branch-based accounting systems. These systems of record were never designed for 24/7 availability, omnichannel experiences, or distributed product sourcing. They struggle with scalability, context, and flexibility.
Systems of interaction, meanwhile, evolve at breakneck speed. From web and mobile to voice, chat, and now intelligent agents, channels are multiplying and fragmenting. No one channel can, or should, hold the business logic.
That leaves a void. And this is where we see the system of intelligence enter:
- A mediation layer that scales interactions without overwhelming systems of record.
- An orchestration engine that aggregates data across multiple systems.
- A context layer that adapts to new channels and enables embedded finance.
“The advent of Generative AI underlines the value of systems of intelligence — orchestrating across multiple datasets and workflows — to translate user intent into results.”
Ben Robinson
From our perspective, this isn’t a future vision, it’s happening now. Our clients, who range from private banks to global insurers, are already using orchestration to connect fragmented systems, launch new propositions, and embed financial services into third-party ecosystems.
Scalability, experience, and new business models
Ben goes on to identify three drivers for systems of intelligence: scalability, customer experience, and business model innovation. Let’s take a moment to unpack each:
Scalability
Traditional systems of record don’t distinguish between transactions and interactions. That creates unnecessary load. A balance enquiry doesn’t need to hit the core ledger in real time, but many legacy architectures treat it as if it does.
Systems of intelligence can do the triage. Cache data for high-frequency, low-value interactions and reserve systems of record for state changes. The result: more scale at lower cost.
Customer experience
True omnichannel requires some context. Pulling portfolio data from one system, risk profile from another, preferences from a third, and doing so in real time. Neither systems of record nor systems of interaction are suited to do this. Systems of intelligence fill that gap, enabling seamless, personalized, and even more importantlty – compliant, experiences.
Business models
Embedded finance, multi-sourcing of products, and distribution through non-traditional channels all require orchestration. If a bank wants to offer a third-party lending product through an e-commerce site, it needs an intermediary layer that can manage product catalogs, eligibility checks, and fulfillment.
In short: scalability keeps the lights on, experience wins customers, and orchestration opens up entirely new revenue streams.
Generative AI: The catalyst, not the replacement
A major theme in the article is the impact of Generative AI. Some argue that GenAI itself could replace the need for a system of intelligence, but Ben proposes the opposite viewpoint — and we agree.
Generative AI lowers the cost of “intelligence.” It enables new natural language interfaces, faster software cycles, and agent-like user experiences. But precisely because channels are proliferating and expectations are rising, orchestration becomes more important.
Without a system of intelligence:
- GenAI risks becoming chrome paint on a rusting car.
- Institutions struggle to integrate multiple models (deterministic and probabilistic).
- Compliance, auditability, and explainability suffer.
With a system of intelligence:
- AI agents can be supervised, coordinated, and contextualized.
- Proprietary data can be harnessed responsibly.
- New interfaces (voice, chat, agents) can plug into consistent business logic.
“GenAI strengthens the case for systems of intelligence.”
Ben Robinson
Embedded finance: Pressure on architecture
In the article, Ben also notes that embedded finance has moved beyond the hype and into structural growth. Valuations have cooled, but adoption hasn’t slowed down. The premise – adapting financial services to context and point of need, is still too compelling.
For incumbents, this introduces pressure. Legacy cores assumed that institutions would distribute only their own products, through their own channels, during business hours. Embedded finance requires the opposite:
- Multi-product, multi-sourced catalogs.
- Distribution through third-party channels.
- Real-time interaction and fulfillment.
Again, systems of intelligence are the enabler. They mediate between internal systems of record, external product manufacturers, and external distribution partners. Without this, incumbents can’t compete with digital-native platforms that were born to aggregate and orchestrate.
Why adoption has been slow
If the case is so compelling, why hasn’t every bank deployed a system of intelligence? Ben highlights four reasons, and from our vantage point we’d add a fifth.
- Platforms take time to mature.
Orchestration layers are not point solutions. They are platforms that need to sit at the center of an ecosystem, connecting systems of record, distribution channels, and external partners. That means value grows as the ecosystem grows, which takes time. In 2019, many of the necessary APIs and integrations didn’t exist. Today, they do. - Market share erosion has been gradual.
Incumbents have been losing ground to digital-native competitors, but the erosion has been steady rather than sudden. Without a burning platform, many banks felt able to defer hard decisions on architecture. That window is now closing as challengers reach scale and embedded finance accelerates. - Rising interest margins reduced urgency.
For much of the last three years, high net interest margins gave banks breathing space. Profits were strong, reducing the pressure to cut costs or modernize systems. But with competition intensifying and rates now shifting, that cushion is thinning. Efficiency gains are no longer optional. - COVID diverted focus and resources.
The pandemic forced institutions to prioritize digital channels above all else, just to keep serving customers. That absorbed resources, delaying investment in the deeper layers of infrastructure. Many institutions emerged with better front ends, but with the same fragile back ends. - And our added point – Perception of complexity.
Perhaps most importantly, orchestration has often been misunderstood as another massive IT program, a multi-year core replacement. In reality, it’s the opposite. A system of intelligence makes it possible to innovate without replacing the core, freeing institutions from the constraints of legacy while extracting more value from existing systems.
additiv’s perspective: Orchestration in practice
Ben highlights additiv as a case in point: A platform designed as a headless orchestration layer. Initially deployed in private banking, we quickly proved our platform to be adaptable to embedded finance, and now to add AI-enabled automation.
Some examples we see in practice:
- Wealth and pensions: Orchestrating between CRM, portfolio management, and risk systems to deliver personalized advice at scale.
- Insurance: Coordinating AI agents to automate claims, from damage detection to policy validation to settlement.
- Credit: Automating mortgage journeys end to end, integrating brokers, valuers, credit bureaus, and lenders.
- Operations & compliance: Embedding audit trails and explainability into every workflow.
In each case, the common thread is orchestration. AI agents, data sources, and workflows only deliver value when connected and contextualized.
Optionality: Future-proofing with systems of intelligence
Perhaps the articles most important insight is about optionality. Institutions are overwhelmed by GenAI hype. They face endless prototypes, uncertain ROI, and integration challenges.
A system of intelligence reduces that burden by:
- Providing a single orchestration layer for multiple models.
- Enabling incremental adoption of use cases.
- Allowing new channels and products to plug into consistent business logic.
In other words: point solutions might deliver productivity gains; systems of intelligence enable transformation.
What comes next
Six years after the original thesis, the case for systems of intelligence has only grown stronger. Scalability, customer experience, embedded finance, and AI all point in the same direction: the need for orchestration.
For incumbents, the question is no longer whether to adopt systems of intelligence, but how quickly. Delay risks ceding ground to digital-native competitors; action creates the foundation for sustainable innovation.
At additiv, we see every day how orchestration turns ambition into execution. As the article concludes:
“This infrastructure is systems of intelligence and their day has come.”
The orchestration imperative
If systems of intelligence are the missing layer in financial services, the next question is simple: is your institution ready to adopt one?
To help you assess, we’ve created a straightforward Orchestration Readiness Checklist. It isn’t a scorecard, but a set of yes/no questions designed to reveal whether your foundations are solid, and where orchestration could make the biggest impact.
Ask yourself:
- Data foundations
Do you have unified, well-governed data that can be accessed across multiple systems of record without manual workarounds? - Process automation
Are your most critical workflows, onboarding, servicing, compliance, already automated end-to-end, or are they still dependent on manual steps? - Channel consistency
Can you deliver a seamless, consistent experience across all client channels, even as new ones (e.g. chat, voice, agents) emerge? - Regulatory posture
Can you demonstrate explainability, auditability, and compliance across every client interaction if regulators asked tomorrow? - Business ambition
Beyond cost-cutting, are you set up to expand distribution, embed third-party products, and create new revenue streams?
If you answered “not yet” to several of these, you’re not alone. Most institutions are on the same journey. The important step is knowing where you stand — and that’s where AI Studio and additiv’s orchestration expertise can help you move forward with clarity and confidence.
Get in touch to learn more.
