Research
Sovereign Digital Resilience — a paper series published under Robotiqa Technologies Ltd.
ORCID iD 0009-0003-1677-684X (opens in a new tab) · Google Scholar · SSRN · Zenodo · Author affiliation: Robotiqa Technologies Ltd.
Independent practitioner. Robotiqa Technologies Ltd. is the author's incorporated consulting practice and research affiliation; it is not a university or academic appointment.
The series examines what it takes for a system of public record to remain trustworthy across decades: where sovereignty actually resides, why concentration in a vendor is a sovereignty problem rather than a procurement one, and what accountability means once decisions are made partly by machines. Every paper is archived with a permanent DOI. Full text is published at sanjeetkumar.com (opens in a new tab); this page is the series index.
The AI operating model
what changes when delivery compresses
-
Sovereign Digital Resilience · May 2026
The AI Gateway Advantage (opens in a new tab)
Value over Spend. Smart Work over Hard Work.
A practitioner’s operating model for scaling enterprise AI without accumulating lock-in. The paper argues that token spend is the wrong governing metric, that an organisation-owned gateway converts model telemetry into a measure of delivered value, and that exit cost is an engineering variable to be driven toward zero rather than a contractual afterthought. Sovereignty, on this account, is an enforced routing policy — not a paragraph in a vendor agreement.
https://doi.org/10.5281/zenodo.21970942 (opens in a new tab)
-
Sovereign Digital Resilience · July 2026
Governing Agentic AI (opens in a new tab)
In Systems of Public Record — Governance Is the Accelerant
Agentic systems acting on records of public consequence present a threat model unlike conventional automation: the actor is capable, fast, and only partly predictable. The paper offers four governing principles, an autonomy ladder describing graduated levels of machine discretion, and the control plane required to enforce them. Its central claim runs against instinct — well-specified guardrails are what allow an institution to adopt agentic AI quickly, rather than what slows it down.
https://doi.org/10.5281/zenodo.21971009 (opens in a new tab)
-
Sovereign Digital Resilience · August 2026
The Intent Bottleneck (opens in a new tab)
When AI Compresses Delivery, Strategy Becomes the Constraint
AI-native delivery compresses construction — the last of six stages between an institution’s obligations and running software. By Amdahl’s law, total gain is bounded by the five stages it does not touch, and cheap construction on an undisciplined intent pipeline simply produces the wrong things faster. The paper sets out an operating model for the upper chain and introduces intent provenance: an unbroken line from a stated obligation to the test that proves it in production.
https://doi.org/10.5281/zenodo.21943397 (opens in a new tab)
Long-horizon accountability
what must still be answerable later
-
Sovereign Digital Resilience · August 2026
Decision Custody (opens in a new tab)
Who Answers for a Decision After the Machine That Made It Is Gone
Automated decisions outlive the systems that produce them. Models are retired, vendors are replaced, platforms are decommissioned — and the decisions remain, still binding on the people they affected. The paper defines decision custody: the obligation to answer for a decision after its originating machinery no longer exists, and the service covenant that carries it. It argues that accountability frameworks track approvals and audit trails while leaving unmanaged the clock that runs after the system is switched off.
https://doi.org/10.5281/zenodo.21959271 (opens in a new tab)
-
Sovereign Digital Resilience · August 2026
The Cold Standard (opens in a new tab)
A Fixed Reference for a Record That Won’t Stop Moving
Records in continuous machine-assisted use drift — not through corruption, but through accumulated ordinary revision. The paper proposes a cold standard: a sealed, human-gated reference against which a working record can be measured, together with an isolation ladder describing degrees of separation from live systems. It is framed as an epistemic instrument rather than a disaster-recovery backup. The question it answers is not how to restore a record, but how to know whether the record still says what it once said.
https://doi.org/10.5281/zenodo.22073326 (opens in a new tab)
Foundations
what long-lived systems owe the record
Short practitioner papers. References are being added on the next revision of Century-Scale Systems and Tokenized Title.
-
Sovereign Digital Resilience · July 2026
Century-Scale Systems (opens in a new tab)
What Land Registries Teach Us About Data Sovereignty
Most technology is designed around a product lifecycle. A land registry is designed around a civilisation lifecycle, and the difference changes every architectural assumption. Five lessons follow from systems built to that horizon: that sovereignty is a property of the record rather than of the data centre holding it, that every dependency is a future liability, and that the actual product of a registry is not data but trust — which is lost discontinuously and regained slowly.
https://doi.org/10.5281/zenodo.21970840 (opens in a new tab)
-
Sovereign Digital Resilience · July 2026
Exit-Capable by Design (opens in a new tab)
Vendor Concentration Is a Sovereignty Risk
De-perimeterisation taught the industry to stop trusting the network boundary. Its modern corollary is to stop trusting the commercial one. When a critical platform can only be operated by a single supplier, the institution has transferred a sovereign function to a private balance sheet — a sovereignty problem wearing a procurement badge. The paper treats exit-capability as a tested design requirement with measurable cost, rather than a clause nobody has ever exercised.
https://doi.org/10.5281/zenodo.21970869 (opens in a new tab)
-
Sovereign Digital Resilience · July 2026
Tokenized Title (opens in a new tab)
Five Governance Questions Every On-Chain Land Registry Must Answer
Dubai has placed property title on-chain at scale, with tokenized deeds, a regulated secondary market, and what the Dubai Land Department describes as the first Property Token Ownership Certificate. The paper poses five questions that any on-chain registry must be able to answer in writing — concerning finality, reversal, the relationship between token and legal title, custody of the authoritative record, and what happens when the chain and the register disagree. Written from the perspective of registry architecture rather than distributed-ledger advocacy.
https://doi.org/10.5281/zenodo.21970983 (opens in a new tab)
Papers are published open access and archived on Zenodo, and are progressively deposited to additional repositories; links appear here as they become available.
Advisory work derived from this series: /advisory