The Foundation: Research & Decision Engineering™
My work has centred on a single structural question:
How do critical institutions - particularly banking and healthcare - transition from traditional operating models to orchestrated, programmable systems without weakening stewardship?
This is not a discussion of technology adoption. It is a re-examination of risk, accountability, and fiduciary responsibility in an environment where decisions are increasingly automated.
The foundation rests on two pillars: forensic research and practical application. Through institutional advisory work and the development of the Decision Integrity Chain™ (DIC™), I focus on redesigning decision architecture so that optimisation does not outrun governance.
These papers extend that effort. They move beyond policy rhetoric into what I describe as Decision Engineering™ - the discipline of embedding purpose, judgment, and accountability directly into execution pathways.
Across domains - from behavioural drift in digital capital markets to operating model fragmentation in healthcare - the same structural risk appears: a widening gap between algorithmic optimisation and human stewardship.
The work is not about resisting automation. It is about ensuring that automation remains accountable.
I. Research: Diagnosing the Drift
The following papers establish the theoretical and forensic basis for the Decision Integrity framework.
137 Countries Are Building CBDCs. Three Things Most Are Not Designing For
The Premise: 137 countries. 98% of global GDP. 49 live pilots. And the most important number in the space is not any of those — it is 98.5%. That is the share of Nigeria’s eNaira wallets that sat unused one year after launch. Full infrastructure. Sovereign backing. Nobody showed up. Not a technology failure — the eNaira worked. A governance failure. The system could not tell a citizen why they should use it, what it would do with their money, or what would happen if something went wrong. When adoption stalled, the government engineered a cash shortage to force behaviour. What followed was not adoption. It was protests.
The 2026 Context: 49 programmes are currently running pilots. Most are having the same conversation Nigeria was having — about ledgers, settlement, and token design. Three questions are not being asked: Why would someone use this instead of what they already have? What happens when the system makes a decision the citizen cannot understand or challenge? And who exactly is this being built for — because in most pilot markets, the unbanked majority is an afterthought, not the design centre.
This paper applies the Decision Integrity Chain™ as both diagnostic and specification, examining Nigeria’s eNaira layer by layer. What that produces is not a post-mortem. It is a map of exactly where the next 49 programmes will break if the architecture conversation does not change. The window for getting this right is not permanent. The evidence for what goes wrong when it closes already exists.
The Fiduciary Gap in AI-Driven Financial Institutions
The Premise: The original paper. Defines the Fiduciary Gap - the structural distance between what an autonomous system is optimised toward and what the institution is obligated to protect. The gap is not caused by failure. It opens because institutional intent changes faster than the parameters encoding that intent. This is the foundational concept from which the Decision Integrity Chain™ was built.
The 2026 Context: Every subsequent paper, advisory engagement, and issue of Decision Engineering™ traces back to this concept. The Fiduciary Gap is not a theoretical risk. It is the current operating condition of most institutions that have deployed AI in consequential decision pathways. This paper names it, defines it, and begins the architecture for closing it.
Decision Integrity in Agentic Retail Banking
The Premise: Examines the paradigm shift from automated AI to “Agentic Banking,” in which autonomous systems are responsible for fundamental financial activities. It describes the Fiduciary Gap, the essential disconnect between algorithmic optimization and the qualitative governance obligations of an institution. Using a case study of the deposit optimization agent “Opti,” the paper illustrates how rule-based implementation can paradoxically drive liquidity risk and destroy economic partnerships.
The 2026 Context: Offers a real-world governance model for programmable institutions. It presents the Decision Integrity Chain™ for decision-level monitoring and outlines the execution guardrails, such as the Fiduciary Hurdle Rate and Delegation Contract. The work defines Decision Engineering™ as a required field of study to ensure that, as AI penetration (ω) grows, fiduciary obligations are encoded in the operating model code.
Reimagining the Future of Banking in a Protocol-Driven World
The Premise: The paper discusses the structural development of financial services from static, platform-based service providers to dynamic, programmable financial institutions.
The 2026 Context: It sets out the architectural imperative for “Fiduciary Boundaries” – those design features necessary to ensure that, as financial institutions become programmable, they remain governed by stewardship rather than algorithmic efficiency
Bridging Fragmented Health Systems in the Age of Subscription, Orchestration, and Ambient Care
The Premise: Examines the pernicious “split” in modern health systems, where two incompatible operating models coexist: one digital and continuous, the other analog and episodic, and in the process, unintentionally create institutional inequality and decision drift.
The 2026 Context: Offers a six-part leadership diagnostic to redirect investment from physical growth to “Decision Integrity” and “Orchestration,” so that value-based care is not just a policy promise, but is engineered into the operating model.
When Stable Deposits Stop Being Stable: A Behavioral Reassessment of Core Funding in Regional Banking
The Premise: Recognizes a new risk in the banking structure post-2023: “Behavioral Drift.” Shows how treasury automation, API connectivity, and yield transparency have turned passive SME and corporate deposits into high-velocity and rate-sensitive liquidity pools.
The 2026 Context: Presents a new stability framework that reconsiders “core deposits” based on observed digital behavior rather than product categorization. Offers boards three key metrics to avoid “Decision Drift” in loan pricing and capital allocation due to the unseen shortening of deposit duration.
II. From Research to Practice
The papers above are the diagnosis.
Decision Engineering™ Weekly is what comes next.
Each issue takes one layer of the Decision Integrity Chain™ and runs it through a real institutional failure — financial services or healthcare — to show exactly where the chain broke and what the structural fix looks like. Eight layers. Eight issues. The research distilled into something you can take into your next governance review, risk committee, or board paper.
The full framework, Decision Autopsies, and all published issues are at Decision Engineering™ — lumathink.com.
If you work in banking or healthcare and are dealing with a decision your institution can’t fully explain, read How I Work.
That’s normally where the conversation starts.
III. Why This Matters Today
As AI becomes autonomous, governance cannot remain outside execution. Governance has to become part of execution.
In all jurisdictions, inflection points in regulation point to one common principle: decisions need to be traceable, bounded, and accountable.
Decision architecture is no longer a choice. It’s a fiduciary imperative.
© 2026 Deepak Aggarwal. All Rights Reserved. Decision Engineering™, Decision Integrity Chain™, DIC™, and DIC ChainTrace™ are trademarks of Deepak Aggarwal. Unauthorized use or reproduction without express written permission is prohibited.