Priyadarshi Pany is Managing Director and CEO of CSM Technologies. He talks about why large government programmes fail and how AI, digital identity and shared digital infrastructure can make services work for citizens. He also explains where humans must stay in charge.
Where do large government digital programmes most often go wrong and how do you de-risk them at the design stage?
Large government technology programmes usually falter when technology is treated as the destination, not as an instrument of institutional reform. The warning signs are familiar. They include unclear ownership, fragmented data, weak interoperability, insufficient field consultation, inadequate cybersecurity and systems designed around departments rather than citizens.
We de-risk projects by starting with policy and process architecture before writing code. We map stakeholders, decision rights, data flows, legacy dependencies, regulatory obligations and citizen journeys. We then design modular platforms with clear APIs, security controls, auditability and measurable outcomes.
Equally important is change management. A technically brilliant platform that officials cannot adopt or citizens cannot navigate is not a successful digital transformation. At CSM, nearly three decades of GovTech experience has taught us that sustainable transformation happens when technology, policy, people and processes are designed as one system.
How are AI, analytics, cloud and cybersecurity changing decision-making and service delivery in Indian governance, and which initiatives show this best?
The shift is from digitising transactions to making government more intelligent, predictive and responsive. Analytics turns administrative data into decision intelligence. Cloud enables elastic infrastructure. Cybersecurity becomes foundational to citizen trust. AI can identify patterns that humans may otherwise miss.
We are seeing this convergence in initiatives such as IndiaAI, BHASHINI and the Ayushman Bharat Digital Mission. At the state level, Bihar’s AI-enabled project monitoring system, developed by CSM, uses milestone tracking, geo-tagged photographs, dashboards and real-time information to strengthen infrastructure oversight. IndiaAI is also building national capabilities around compute, datasets, innovation and safe AI.
The larger lesson is that AI should not sit on top of weak digital foundations. It becomes transformative when connected to reliable data, interoperable systems, secure infrastructure and clearly defined public outcomes.
With MOSIP now part of CSM’s digital identity work, how does a trusted digital ID become the foundation that lets benefits, land and health records work together across departments?
A trusted digital identity is the connective tissue of DPI. Without a reliable way of establishing who a person is, government databases remain islands, even when they are digital.
MOSIP’s open-source, modular and vendor-neutral architecture gives governments flexibility to build foundational identity systems suited to their policy and institutional contexts. That identity layer can then connect, with appropriate consent and governance, to social protection, healthcare, education, taxation, financial inclusion and land services. One trusted identity can reduce repetitive verification, improve targeting of benefits and make services more portable across departments.
CSM’s engagement with MOSIP strengthens our ability to integrate identity with the broader ecosystems we already work across. With billions of people globally still lacking digital IDs usable for online transactions, the priority is not simply identification. It is trusted, interoperable and citizen-controlled identity infrastructure.
CSM offers agentic AI and AI Foundry. Where can AI agents help in government workflows today, and where must humans stay in charge?
AI agents are extremely valuable where government workflows are repetitive, rules-driven and information-intensive. They can triage grievances, extract information from documents, prepare case summaries, monitor compliance and generate reports. They can also detect anomalies, assist officials with knowledge retrieval and coordinate multi-step administrative workflows.
CSM AI Foundry, for instance, combines agents with RAG (Retrieval-Augmented Generation), model management, observability, explainability and responsible-AI guardrails.
But government is fundamentally about rights, entitlements and public accountability. I would never advocate handing final decisions on welfare eligibility, land ownership, taxation disputes, policing or healthcare outcomes entirely to an autonomous system. Humans must retain authority wherever decisions materially affect rights, livelihoods or access to public services.
The right model is not AI replaces the official. It is AI amplifies the official while making the decision more informed, transparent and auditable.
Which sector (mining, agriculture, healthcare, education, land) is the hardest to digitise, and why?
I would choose land. Mining, healthcare, education and agriculture are all enormously complex. But land combines historical records, legal rights, physical geography, multiple authorities, disputed ownership, legacy documentation and highly consequential human interests.
A digitised land record is not automatically a trusted land record. The underlying data must be accurate, legally recognised, spatially consistent and continuously updated. That is why land digitisation needs much more than a portal. It needs cadastral integration, GIS, workflow automation, document management, verification mechanisms and institutional coordination. Our experience with solutions such as LANDMAS reinforces this point.
The deeper challenge is that land is an economic asset, a legal instrument and a social reality at the same time. Technology can create transparency and efficiency. But only sound institutional processes can convert digitised information into trusted property rights.
How do you design digital platforms so they work for citizens with low digital literacy?
The first principle is to design for the citizen, not for the software architecture. If a citizen needs a training manual to access a public service, we have probably designed the system incorrectly. Interfaces should be multilingual, mobile-first, accessible, intuitive and increasingly voice-enabled.
Assisted access through Common Service Centres and frontline government workers remains equally important. Digital inclusion cannot mean digital-only. India’s BHASHINI is particularly significant because language itself can be a barrier to digital participation. Its ecosystem now supports more than 34 languages, 22+ language services and hundreds of AI models.
We also need to design for intermittent connectivity, low-end devices and users unfamiliar with digital terminology. Ultimately, inclusion is measured not by how sophisticated the platform is, but by whether the least digitally confident citizen can complete a task independently, safely and with dignity.
How can separate departmental systems be connected so citizens deal with one government and not many portals?
The answer is interoperability by design. Departments do not need one giant monolithic system. They need common standards, shared identity and consent mechanisms, interoperable registries, APIs, secure data exchange and clearly defined governance for who can access what data and why. This allows specialised departmental systems to remain fit for purpose while giving citizens a unified service experience.
India’s DPI journey offers a powerful example. Foundational platforms can become shared rails on which multiple services operate. The Ayushman Bharat Digital Mission, for instance, uses registries and interoperable digital health infrastructure to connect citizens, professionals, facilities and health records.
At CSM, our approach is similar. We connect ecosystems rather than merely screens. The citizen should experience one government, even if several departments and technology platforms work behind the scenes.
As governments digitise faster, where are the biggest opportunities for technology companies and enterprises to partner with them?
The largest opportunity is moving beyond application development into building the digital infrastructure that governments increasingly need. That covers identity, interoperability, data platforms, cybersecurity, AI governance, cloud, analytics, GIS and sector-specific DPI.
Emerging economies are particularly interesting. Many are not burdened by decades of fragmented legacy architecture and can leapfrog directly to modular, interoperable systems. Africa is a strong example. Kenya’s digital mineral royalty system shows how a sector-specific platform can improve compliance, revenue visibility and accountability across a national ecosystem.
For technology companies, this opportunity comes with a responsibility. Governments are not ordinary enterprise clients. Their systems affect citizens at population scale. The winning partnerships will combine engineering capability with policy understanding, domain expertise, local capacity building, cybersecurity and long-term institutional ownership. The future belongs to companies that can build public digital infrastructure, not just software.
What will a truly smart government look like when AI and Digital Public Infrastructure work together, and what is the first thing governments should get right?
A smart government will not simply have more AI. It will have better institutional intelligence. Citizens should be able to access services proactively. Officials should see emerging problems before they become crises. Policy decisions should increasingly be informed by real-time evidence, not fragmented reports. India already shows the potential of this model. As of July 2026, the country had more than 144 crore Aadhaar IDs, 55.49 crore UPI users and 71.66 crore DigiLocker users.
The first thing governments must get right is the foundation. That means trusted data, interoperable architecture, cybersecurity, privacy and clear accountability. If the foundation is weak, AI merely accelerates bad decisions. If it is strong, AI can turn DPI into an intelligent public-service engine.
