MENU
Manyank Manish examines why India's public finance system is built for accounting and control rather than planning, and argues that AI can help close the data gaps that force policymakers into reactive, ad hoc decisions.
Mayank Manish is a Public Policy professional with 8+ years of experience. He works to translate complex data into actionable policy & insights to architect more effective systems. His expertise spans across Building State capacity, Al Safety & Governance, Data & Tech policy, Public Finance Management (PFM), Climate Action, Political economy, Public health, and Urban Development. He currently works as a PFM consultant with eGov Foundation.
Public finance management in India, as Manyank Manish describes it, was never designed to help decision makers plan. It was built for accounting, control, and auditing. This design choice has real consequences: district officials, state governments, and national policymakers lack the real-time, structured data they need to anticipate problems rather than just respond to them once they happen.
Manish walks through how this plays out in practice, using the example of flood preparedness in Assam to show how the absence of timely, visible data prevents proper planning at both the district and state level. He then traces the downstream effect of this data fragmentation: without evidence to guide decisions, resource allocation tends to favor whoever advocates loudest, rather than whoever has the greatest actual need.
The second half of the talk turns to what can be done about it. Manish lays out a vision centered on standardized, public datasets validated through ground truthing, and argues that AI has a specific and useful role to play here, not in making decisions, but in doing the unglamorous work of structuring, collating, and cleaning data so that researchers, policymakers, and experts can focus on inference and action.
Manish, Manyank. "Data-driven Decision making for fiscal allocation." Episode 84 of Big Ideas. XKDR Forum, 2nd November 2026. Video, 3:31. https://www.xkdr.org/viewpoints/data-driven-decision-making-for-fiscal-allocation-big-ideas-ep-84
Manyank Manish opens by naming the structural problem at the heart of Indian public finance: the system was designed around accounting and control, auditing and compliance, not around giving decision makers the tools to plan. This is not a minor technical gap. It means that people sitting at the district level, the state level, or the national level cannot actually use the system to make better planning decisions, even when they want to.
The reason traces back to how data moves, or fails to move, through the system.
"The data which exists definitely exists in silos, in fragmented forms. It doesn't reach to them in time. There is no real time visibility in a lot of variables. So they can't plan."
This lack of real-time visibility is not an abstract complaint. It directly prevents officials from acting ahead of problems instead of scrambling after them.
Manish grounds this in a concrete example: a district in Assam cannot plan for floods in advance because it does not have visibility into what all is getting affected. The absence of timely, structured information means disaster preparedness becomes guesswork rather than planning.
The same problem repeats at the state level, but with a different shape. State governments face three separate visibility gaps when trying to decide which districts to prioritize. They do not know which districts are more likely to face the impact of disasters. They do not know which districts already have provisioning left over from previous years. And they do not know which districts need which specific kind of support.
Manish points out that needs can vary significantly even among districts facing similar broad risks:
"One district might require help in repairing an embankment or a dam. Another district might require help with their electricity system or roads, and so on."
Without data that captures this level of specificity, state-level prioritization cannot be precise, no matter how well-intentioned the decision makers are.
These data gaps do more than slow things down. Manish argues they actively distort how resources get allocated. In the absence of evidence showing where the need actually lies, decision making defaults to a different, less fair mechanism.
"Rather than seeing where the evidence leads, where the data leads, what is the actual need of people, districts, departments, what ends up happening is that whoever has the loudest voice usually gets the most money, usually gets the most preference and usually gets the most priority."
This is the central cost of fragmented, delayed data: it does not just create inefficiency, it creates inequity in how public resources get distributed, since need and advocacy strength are not the same thing.
Manish's proposed fix starts with building systems that define what is actually needed, and then validating those definitions against reality through ground truthing with the people who exist on the ground. The foundational step underneath this is building datasets that are standardized, so that data from different geographical entities, different states, different districts, can actually be compared against each other.
Making this data public is the next piece. If broad trends are visible to outside observers, it becomes possible to see where governments are currently focusing their attention and resources, and to compare that against where the evidence suggests the focus should be.
Manish is specific about what role he sees AI playing in this process. It is not about AI making the decisions. It is about AI handling the groundwork that currently consumes time and capacity: standardizing data structures, collating fragmented sources, and processing and cleaning the data.
"AI can help in so that the people, the researchers, the experts, the academics, the policy makers can then just look at this data, form their inferences, take the decisions and make better plans."
The intended outcome of all this is a shift in the nature of policy making itself, away from reaction and toward preparation.
"So that policy making and decision making is not as ad hoc as it currently is and is more planned, not more reactive, but rather more in fact prepared for it."
Manish closes by stating his hope plainly: that technology and AI get used to build decision making on evidence, and that policy making gets the time to prepare for crises rather than simply reacting to them one after another.
The complete transcript file is available to download below.
19 Oct 2026
Arbind Modi
05 Oct 2026
Abhishek Anand
21 Sep 2026
Sanjiv Shah
2026, XKDR | GSTIN: 27AAACX3640C1ZN
by Sane Difference
