Government software has a harder brief than any commercial product, and it is not the scale. It is that the user cannot choose a competitor, cannot be assumed to own a smartphone or read English, and cannot be turned away because their case is inconvenient. A citizen service portal that works beautifully for the confident urban user and fails the elderly pensioner in a Tier-3 town has not succeeded — it has simply moved the exclusion online. Everything about how we build for the public sector follows from that.

Identity, payment and documents are national rails now

India has built genuinely world-class public digital infrastructure, and the job is to use it correctly rather than reinvent it. Aadhaar-based authentication and e-KYC for identity, with informed consent and the privacy safeguards the law requires. DigiLocker so a citizen presents an issued certificate rather than uploading a photocopy nobody can verify. UPI for fee collection, because it is the payment rail every citizen already trusts. API Setu for department-to-department data exchange, so a citizen is not asked for a document one arm of government already holds. Used well, these turn a twelve-document, three-visit process into a single online form. Used carelessly, they become a new single point of failure, so we build every one of them with graceful fallback for the citizen whose Aadhaar authentication fails on the day.

Accessibility is a legal and moral requirement, not a nice-to-have

The Guidelines for Indian Government Websites and the RPwD Act make accessibility a compliance obligation, and WCAG conformance is the measurable standard. But real accessibility in India goes further than a screen-reader pass: multilingual interfaces so the citizen reads the form in their own language, not just English or Hindi; interfaces that work on a low-end phone over a 2G connection; readable typography and clear flows for users who are not digitally confident; and an assisted-service path — a Common Service Centre operator or a help desk — for the citizen who genuinely cannot self-serve. We build to WCAG and GIGW from the first sprint, because retrofitting accessibility is both more expensive and, done late, never quite honest.

Grievance redressal and workflow are the actual product

Most e-governance is, underneath, a workflow: an application is submitted, routed to the right officer, moved through defined stages with defined timelines, escalated if it stalls, and closed with an outcome the citizen can see. Grievance redressal is the same shape, and it is where public trust is most directly won or lost. We build workflow engines where the process is configured rather than hard-coded — because government processes change with every notification — with SLA clocks on each stage, automatic escalation when a stage overruns, and full visibility to the citizen so they are never left wondering whether their file is moving. A grievance that disappears into an inbox is worse than no portal at all.

Audit trails, RTI and the assumption of scrutiny

Public systems are built for scrutiny in a way private ones rarely are. Every decision, every state change, every access to a citizen's record has to be logged in an append-only trail that an official with system access cannot quietly alter, because that trail is what answers an RTI request, a CAG audit, or a court. RTI itself is a workflow with statutory timelines and defined appeal stages, and it deserves to be built as one rather than tracked in a register. Data protection under the DPDP Act applies to government too, with its own carve-outs, and role-based access with the principle of least privilege is not optional when the data is every citizen's. We build assuming the system will one day be audited line by line — because sooner or later, it will be.