Sangeeth Manoharan

‘Advanced AI systems can be applied responsibly, transparently, and at scale within highly regulated environments without compromising governance or accountability’

Part of the AI Leadership Index – a curated list of the most inspirational figures shaping the AI revolution.

Sangeeth Manoharan,Head of Artificial Intelligence at Marston Holdings, is an AI and Automation leader focused on transforming intelligent operations across complex, regulated environments. With experience spanning Judicial Services, Civil Enforcement, Local Government, Finance, Audit, and Customer Contact Centres, he has delivered award winning enterprise-scale AI solutions that improve efficiency, strengthen service delivery, and enhance citizen engagement.

His approach combines strategic vision with hands-on execution. He specialises in embedding AI into high-volume operational environments where compliance, auditability, and service continuity are critical. Rather than pursuing innovation for its own sake, he focuses on deploying fit-for-purpose AI systems that deliver measurable impact while maintaining transparency and accountability.

Alongside implementation, Sangeeth contributes to wider discussions on AI standards, policy, and governance. Through cross-sector collaboration, he helps translate emerging regulatory and ethical principles into practical, enterprise-ready frameworks that support responsible adoption at scale.

He positions generative AI and intelligent automation as enablers of sustainable growth, ensuring that technological advancement aligns with organisational objectives and public trust. His work reflects a belief that successful AI leadership is not about adopting the most powerful models, but about applying them thoughtfully, safely, and in ways that create long-term value within regulated environments.

Please share a case study of how you have you innovated with AI over the last year?

Recently, I led the design and delivery of an in-house AI platform transforming how Penalty Charge Notice (PCN) challenges are managed for local authorities across the UK — one of several AI initiatives I have delivered to modernise high-volume, regulated services.

As enforcement processes have become increasingly digitised, councils have experienced a significant rise in appeals. Reviewing large volumes of unstructured correspondence, including emails, uploaded documents, and supporting evidence, had traditionally been manual, time-consuming, and resource intensive.

I spearheaded the development of an AI platform that combines a citizen-facing interface with a robust backend intelligence engine capable of analysing documents, extracting contextual information, identifying evidential gaps, and supporting structured case assessment. The solution was architected with a human-in-the-loop model to ensure legal compliance, procedural fairness, auditability, and consistent quality assurance.

Drawing on my involvement in cross-industry standards development and national AI policy discussions, I ensured the platform aligned with emerging governance frameworks and best-practice principles from the outset.

Since deployment, the platform has significantly improved assessment consistency and reduced case resolution times while maintaining strict regulatory alignment. Recognised as Highly Commended at the UK IT Industry Awards 2025, the solution is now being extended into other high-volume, regulated services, including complaints and permit processing.

This work demonstrates how advanced AI systems can be applied responsibly, transparently, and at scale within highly regulated environments without compromising governance or accountability.

What is the biggest challenge you have faced with AI innovation and how you have overcome it?

The biggest challenge I have faced in AI innovation has been deploying advanced AI within a highly regulated, high-volume service environment where legal compliance, public trust, and operational continuity are non-negotiable.

The question was never simply “Can we build this?” but rather, “Can we deploy it responsibly, at scale, in a live production environment handling citizen data, legal evidence, and statutory processes?” Applying natural language processing and document intelligence to unstructured inputs from emails, PDFs, images, supporting evidence, required not only technical capability, but full traceability, auditability, and governance alignment. The risk of over-automation or unintended bias was significant, and many off-the-shelf tools lacked the transparency and control required for accountable deployment.

To address this, I ensured governance was embedded from the outset, bringing together legal, operational, and compliance expertise as part of the core design process rather than as a downstream control. The architecture was designed around modularity, human oversight, and clear decision boundaries, positioning AI to augment professional judgement rather than replace it. Explainability and audit trails were treated as foundational features, not optional enhancements.

The result was a stable, scalable solution trusted by stakeholders and capable of operating confidently within regulatory constraints.

This experience reinforced a key leadership lesson: meaningful AI innovation is not about pushing technological limits. It is about applying AI thoughtfully, safely, and pragmatically within real-world constraints, balancing ambition with accountability to deliver sustainable impact.

How do you balance innovation with ethical safeguards?

Innovation without ethics is risk, especially in AI, where decisions can directly impact people's rights, experiences, and trust in systems. I believe ethical safeguards must be designed into the innovation process, not added afterwards. That means involving legal, operational, and user experience experts from day one. It also means defining clear boundaries: what the AI should and should not do, where human oversight is required, and how outcomes can be explained or challenged.

At Marston, we designed our AI to support decision-making, not replace it. It handles evidence gathering, document analysis, and triage but escalates complexity and final decisions to human reviewers. We also prioritised transparency: everything the AI sees and does is auditable, traceable, and reviewable. We use open-source technology intentionally to avoid black-box dependencies and maintain full control over how decisions are made. And we actively test for edge cases and bias to ensure fairness in real world conditions.

Ultimately, I see ethics not as a constraint on innovation, but a framework that enables responsible, scalable, and trusted AI adoption, especially in high-stakes environment

How do you believe the artificial intelligence space will change over the next 12-24 months?

Over the next 12–24 months, AI will move decisively from experimental to embedded, shifting from standalone tools to integrated, everyday infrastructure across both public and private sectors. Agentic AI frameworks will mature rapidly alongside multimodal systems and orchestration layers, enabling AI not just to analyse information, but to initiate, coordinate, and manage end-to-end workflows. This will be particularly transformative in document heavy, regulated environments such as legal, finance, and public services.

We are already seeing early evidence of this shift. The Madras High Court in India has begun using an AI system known as Super Law Courts to assist in reviewing and organising complex arbitration documentation under judicial supervision. This signals that AI when deployed with structured human oversight and procedural safeguards, can operate within even the most tightly regulated judicial settings. As governance models mature, similar controlled adoption is likely to expand globally.

As capability advances, the focus will move from what AI can do to how safely, transparently, and cost-effectively it can do it. Governance, explainability, auditability, and fairness-by-design will become foundational requirements.

Organisations that succeed will combine domain expertise, responsible design, and rigorous safeguards to deliver trusted, fit-for-purpose AI and agentic automation at scale.