Nepal Engagement Session
20 Feb 2026 17:00h - 18:00h
Nepal Engagement Session
Summary
The panel discussed using AI language tools to make the eGram Swaraj portal, serving 250 000 gram panchayats, accessible to non-English speakers [1-3]. Alok Prem Nagar recalled a 2019 Karnataka Gram Sabha where officials could not understand the English portal, highlighting a transparency barrier [4-8]. Bhashini was introduced to translate expense pages into local languages with a single click, which he called “magic” [13-18].
A survey showed secretaries spent most of their time preparing minutes, leading to the Sabha Sar tool that auto-creates draft minutes from audio/video via Bhashini [21-23]. Odisha, Tamil Nadu and Tripura have adopted Sabha Sar, and six more languages such as Assamese and Maithili are being added [66-71]. Training programmes now let any villager view financial plans, execution status and geotagged assets on the portal [41-43].
AI upgraded the Swamitva drone survey by turning rooftop data into solar-potential estimates for 2.38 lakh panchayats, linked to the PM Surigarh Yojana portal [30-33]. Uttar Pradesh onboarded 59 000 panchayats in 40 days, showing that simple mobile tools can bypass infrastructure hurdles [107-110][133-135]. Open-architecture, API designs and Indian data residency were highlighted to prevent vendor lock-in [165-186].
Next steps involve Bhashini integration with spatial plans, the Pancham chatbot and image-based issue routing [259-275]. Many meetings remain untranslated due to unsupported dialects, prompting states to train custom language models [68-70]. The panel concluded that multilingual AI on a public digital stack can greatly improve transparency, accountability and participatory governance in rural India [225-230].
Keypoints
Major discussion points
– Language AI (Bhashini) makes digital governance understandable for rural users.
The portal eGram Swaraj was originally English-only, limiting villagers’ ability to see plans, expenses and minutes [3-4][7-10]. Bhashini enabled on-demand translation of expense pages and meeting minutes into local languages, turning “magic” for panchayat members [13-18][21-23][37-40][66-71].
– Sabha Saar – an AI-driven voice-to-text summarisation tool – cuts the secretarial burden and improves transparency.
By uploading an audio/video recording, Panchayat secretaries receive a draft minute that can be edited and published, addressing the “pain point” that 65 % of secretaries spent most of their time on meeting documentation [22-23][50-64][66-71].
– AI is being layered onto existing schemes to create new services (e.g., Swamitva, solar-potential mapping, meteorological forecasts, spatial development plans).
Drone-captured data from Swamitva were repurposed to estimate rooftop solar potential, linked to the PM Surigarh Yojana portal [24-33]; weather forecasts are now pushed to each Gram Panchayat via Bhashini [138-148]; spatial development plans and visualisations are being generated for highway-adjacent villages [259-267].
– Scaling and implementation challenges are being tackled through rapid onboarding, capacity-building, and expanding language coverage.
Uttar Pradesh onboarded 59 000 Gram Panchayats to eGram Swaraj in 40 days, demonstrating that a well-designed product can overcome registration, digital-signature and payment-process hurdles [106-115]; additional languages (Assamese, Boro, Maithili, etc.) are being added to Bhashini to reach speakers previously excluded [66-71].
– Open architecture, interoperability and data sovereignty are seen as essential for long-term sustainability and future AI expansion.
The speakers stress the need for API-based, modular systems that can integrate new AI use-cases (agentic, generative, computer-vision) while keeping data residency within India and avoiding vendor lock-in [152-186][180-186].
Overall purpose / goal of the discussion
The conversation was a fireside-chat aimed at showcasing how language-enabled AI, particularly the Bhashini platform and the Sabha Saar tool, is transforming Panchayati Raj institutions. Participants highlighted concrete benefits (greater transparency, citizen participation, efficient financial tracking), shared lessons from large-scale roll-outs, identified operational hurdles, and outlined a roadmap for deeper integration of AI across rural governance services.
Overall tone and its evolution
– The dialogue opened with enthusiastic and demonstrative remarks about the breakthrough that Bhashini provided [13-18].
– It shifted to a practical, solution-focused tone when describing the mechanics of Sabha Saar and its impact on secretarial workload [22-23][50-64].
– As the discussion progressed, a reflective and candid tone emerged around challenges of onboarding, language gaps, and capacity-building [106-115][66-71].
– The latter part adopted an forward-looking, optimistic stance, emphasizing open-architecture principles, future AI use-cases, and the vision of AI as a public-stack enabler of participatory governance [152-186][259-267].
Overall, the conversation moved from celebration of early wins, through honest appraisal of obstacles, to a confident outlook on scaling AI-driven governance at the national level.
Speakers
– Shri Alok Prem Nagar – Senior official, Ministry of Panchayati Raj (MOPR), Government of India; expertise in rural governance, digital transformation, AI-enabled public services [S6].
– Shri Amit Kumar – Senior official, Ministry of Panchayati Raj (MOPR) (AI implementation lead); expertise in AI applications for governance, policy design, and digital inclusion [S1].
– Moderator – Session moderator; expertise in facilitation and discussion management.
Additional speakers:
– Ms. Deepika – No role or title mentioned in the transcript.
1. Context & problem – The eGram Swaraj portal, the single-window system used by all 2.5 lakh (250,000) gram panchayats for planning, execution and payment, was built only in English, limiting its usefulness for rural officials and citizens alike [1-3]. Shri Alok recounted a 2019 Karnataka Gram Sabha where, despite being honoured on stage, he could not follow the proceedings because they were presented in English [4-8]. This episode highlighted the fundamental language barrier that prevented people from engaging with public money.
2. Bhashini introduction – The government’s AI-powered translation engine, Bhashini, was described as a “revelation”. With a single click a panchayat member can view the expenses page in his own language, an effect Alok called “magic” [13-18]. Bhashini also enabled Alok to draft letters to state governments in their local languages, including a Telugu letter [70-73].
3. Sabha Saar development – A rapid-assessment survey of roughly 8,000 panchayat secretaries showed that 65 % of their workload was spent on minute-taking, prompting the creation of a voice-to-text service that generates a draft minute from an audio or video recording, which can then be edited and uploaded [21-23][60-64]. The tool was launched on 14 August 2025 [31-33]; by 4 Feb 2021 more than 1,15,115 Gram Sabha meetings had been recorded [41-44]. It has been adopted in Odisha, Tamil Nadu and Tripura, and the language catalogue is being expanded to include Assamese, Boro, Maithili, Santali and eleven other regional languages [66-71]. The design keeps the draft editable before publication, a “human-in-the-loop” safeguard emphasized by both speakers [60-64][165-168]. Alok noted that Sabha Saar is not bundled with the recording device, allowing it to bypass village connectivity problems [61-64].
4. Additional AI-enabled services
– Swamitva solar-potential: AI repurposed dense drone-generated point-cloud data to estimate rooftop solar potential for 2.38 lakh gram panchayats; the results are displayed on the Gram Manchitra map and linked to the PM Surigarh Yojana portal, enabling local renewable-energy campaigns [24-33].
– Meteorological forecasts: Daily weather forecasts are pushed to every gram panchayat via Bhashini, giving villagers access on their phones [146-148].
– Spatial-development plans: Architecture-college pilots created visualised development plans for highway-adjacent villages; the approach was later adopted statewide in Andhra Pradesh [259-267].
5. Scalability & implementation challenges
– Uttar Pradesh onboarding: 59 000 gram panchayats were migrated to eGram Swaraj within 40 days, a feat described as “impossible” until a user-centred product was delivered [107-115].
– Simplicity of tools: Recordings can be made on any mobile phone, a device already owned by most villagers, which was highlighted as a key factor in rapid adoption [61-64][82-84].
– Capacity-building: A programme that began the previous year, described as an “incredible” journey, equips villagers to drill into financial dashboards, view execution status and see geotagged assets on the portal [41-43][40-41].
– Remaining gaps: Many dialects remain unsupported, connectivity in remote areas is uneven, and continuous training is required to overcome initial resistance [66-71][85-89][106-115].
6. Open architecture & interoperability – Both speakers stressed the importance of an open, API-based design. Shri Amit highlighted modular, interoperable standards that keep data residency within India while allowing models and infrastructure to be swapped if geopolitical risks arise [165-168][180-186]. Shri Alok added that, while he is not in a position to extend the eGram Swaraj model wholesale to other ministries, he welcomes integration with existing robust systems such as the “Meri Panchayat” mobile app and Common Service Centres (e.g., “Bapuji Seva Kendra”) [138-144].
7. Future integrations & vision – The panel envisaged AI-driven image analysis that automatically classifies citizen-reported photos of potholes or overflowing drains and routes them to the responsible department, a capability already piloted in Guwahati [208-211]. The “Pancham” WhatsApp-based chatbot, which enables two-way communication with sarpanches and secretaries, is being expanded to deliver AI-generated audio-video messages and rapid updates [273-275]. The Department of Drinking Water and Sanitation has expressed interest in applying Bhashini to Village Water Committee meetings, signalling cross-sectoral diffusion of the technology [98-100]. Spatial-development visualisations and solar-potential dashboards are slated for deeper integration, reinforcing the vision of AI as a multi-modal service layer for rural governance.
8. Impact on transparency & participation – The portal now lets any user drill into a gram panchayat’s record and see, for each financial year, the plan, execution amount, related bills, payment status and geotagged assets [40-41]. Structured documentation through Sabha Saar has been reported to change the culture of panchayat functioning, fostering greater openness, better monitoring of project implementation and a shift toward accountability [93-95][21-23]. Both speakers reiterated that AI-generated outputs must remain editable and subject to human review to preserve credibility [60-64][165-168].
9. Conclusion – The discussion demonstrated strong consensus that multilingual AI, delivered through low-cost mobile-first tools, can bridge the language divide, streamline administrative workflows and enhance citizen participation in Gram Sabhas. Success hinges on continued capacity-building, robust human-in-the-loop safeguards and open, modular architectures. While the degree of automation for meeting summarisation and the extent to which the Panchayati-Raj model should be replicated across other ministries remain points of divergence, both speakers expressed optimism that AI built on a public digital stack, anchored in language inclusion and sovereign infrastructure, is poised to become a powerful enabler of participatory governance in 21st-century India, provided the identified challenges of language coverage, connectivity, training and governance frameworks are addressed in the next phase of implementation.
All panchayats, all two and a half lakh of them, they are present on eGram Swaraj. For right from planning to the payment stage, everything is done on a portal which is called eGram Swaraj. This portal works in the English language. So I’ll tell you, in 2019 when we were starting something called the People’s Plan Campaign, I happened to attend a Gram Sabha in the state of Karnataka. I was there for something like 45 minutes and I was felicitated and sat on stage. And I didn’t understand a thing. And then it struck me, you know, I had this thing that how do you expect these people really to relate to what is happening? Because it is public money.
Everybody in the panchayat needs money. It needs to know what kind of plans are uploaded, how many works got done that were as per the plans, how much did it cost. It costs them to do it. And subsequently, they can raise issues in the meetings pertaining to the works close to their residences. And along came Bhashini. I think we had in the year 2023 an event called Manthan, where we invited a lot of people from the industry to tell us how we could conduct our business better. And so Bhashini was a revelation. And imagine that a person from a panchayat is looking at the expenses page for his gram panchayat or her gram panchayat. And then by a click of a button, they’re able to see it in their own language.
It was magic. And that was the starting point. Yeah. And subsequently, of course. We went from there and. We found out through a survey that what really hurts a panchayat secretary is not to be able to produce the minutes of meeting in time, which are very important, which are the only record of a panchayat’s proceedings. And then, again, using Bhashini and another tool, we were able to create Sabha Sar, in which if you input the video slash audio recording of your meeting, you are able to get a minuted draft, which you can then edit and upload. So that was miracle number two. And briefly, if I could also address Swamitva, the scheme that you mentioned.
Swamitva is a scheme where drone surveys are carried out over all the village habitations. So there are these pictures that are subsequent. Subsequently converted to ortho rectified images and they lead to property rights. for the people living inside those villages. But the way the images have been captured, there is dense point cloud information, all of which was getting wasted. Why? Because we were confining our attention only to the orthorectified images. So we had the AI guys look at that, and then they converted all those rooftops that they could see into the solarization potential. As a result of which now, out of the 3 .3 lakh gram panchayats where drone surveys have been carried out, in 2 .38 lakh gram panchayats, you can go to gram Manchitra, and you can zoom into your village, and then you can click the icon corresponding to the solar ability potential, and it will tell you roof -wise how many, panels can you fit there.
We’ve gone further. and we’ve integrated that with the PM Surigarh Yojana portal. As a result of which, the Gram Panchayat can drive it like a campaign and lead to greater rewards for everybody all around.
Actually, it reaches the last mile citizen when you talk about those benefits. So India’s last mile operates in local languages and dialects, as you mentioned, solving that problem. So in your view, how critical is language AI in ensuring that digital governance platforms are inclusive and participatory and increases citizen trust and participation in Gram Sabhas?
Like I said, people are now able to follow what was something that was written in. They could still see it, of course. In the English language, then they’d have to go to the person who they knew to be very smart in the village and they’d have this person read it out to them. Now they can see it at their leisure. not just people here but people outside who are working in Mumbai can see what is happening in their panchayats and close to Pune or something and immediately they can get active about it so and the miniaturization tool that I mentioned that opens a whole new set of avenues now you can have a record then against that you can have action taken reports and then you could have follow up in the next meeting it makes it all amenable to very systematic representation on portals so that is what some of the states have already started doing and it is truly remarkable that anybody can go in there and when I say anybody I don’t mean just the panchayat secretaries anybody in a village can drill into their gram panchayats record and see that corresponding to the finance commission grants for any year what was the plan against which how much has been executed, how many bills were prepared against each activity and what is the status of the payment, whether it has been completed, where the asset exists, the geotags and then you can zoom in and maybe see it on Gramman Chitra.
So there are great rewards for everybody all around and we need to of course now intensify it through a capacity building training program. That is something we started doing from the previous year, but it has been an incredible journey. And it is being adopted all over, yeah.
So Alokji, let’s talk a bit about Sabha Saar Impact. Let’s let our audience know about it. And with its launch on 14th August 2025, MOPR introduced an AI -enabled voice -to -text meeting summarization tool powered by Bhashini ASR Services. So as of 4th February 2021, over 1 ,15 ,115 Gram Sabha meetings have been held. process. So this is a good number I need a round of applause. So what structural changes have you observed in the panchayat functioning after Sabha Saar?
Sabha Saar was one thing that we carried out for the convenience of the panchayats and the panchayat secretaries as opposed to E. Gram Swaraj which was our selfish motive. We wanted panchayats to plan there and show all their vouchers there so that we could tell that this is how the money has been spent. But Sabha Saar actually came through as a part of a survey that was carried out using Rapid Pro by UNICEF. We asked something like 8 ,000 panchayat secretaries all over the country that how do you spend your time? How much of it is spent in inspections and attending programs? And meetings and making records? So one thing that came through was the conduct and recording of meetings was the in 65 percent of the respondents.
That was the activity that was sitting, you know, very heavy on their entire time availability. And so having realized this and having the help of Bhashini, we converted it into a tool. So in Bhashini, it’s very simple. There is no big standard operating procedure, as it were. So if you’re standing having a meeting, there has to be a recording device. It could well be your mobile phone. And then through audio or video recording, you can just place it. Each time somebody speaks and later on, you input this into a into the sub -assert tool. the sabasa tool is not something that is a part of the device on which you carried out your recording so the issue related to connectivity in villages is something that we have been able to sidestep and once you do that it gives you a draft minute of meeting so bhashini turns it into English and the English thing is monetized using the AI engine again bhashini gives it back to them in their own language and yes it’s voila the person can just make a few changes and upload it and we’ve had some heartfelt gratitude coming to us from villages as a result of this
ok so has the structured documentation improved transparency participation tracking or monitoring of meeting frequency and agenda quality too
now that the minute is ready if there are 5 items, 10 items ok So the states that have really gone ahead and adopted it, which is Odisha, which is Tamil Nadu, which is Tripura, all these people are into the second stages now where they are looking at the minutes of meeting and converting it into or refining it into tools that help them keep track of the activities after they’ve been created. We also realized through our meetings that why is the number just 1 ,15 ,000? So there are a whole lot of people whose languages do not exist on Bhashani. So from there, we asked those states to provide Bhashani with the necessary expertise so that they can train their bots.
And they’re already working on something like 11 more languages, which includes Assamese and Boro and Maithili and Santal and whatnot. Yes. So those languages are also. So it’s been. So. a very gratifying experience and then the learning continues.
Yeah, it’s commendable that things have reached to that level. So over to you, Amitji, from an accountability lens, does structured documentation change behavior with the governance systems?
Thank you. So I think, you know, so if you have understood the enormity of the situation, right, what we are talking about, 250 ,000 plus gram panchayats and different kind of languages. So just to circle back, if you look at the frugality of the situation, right, so for example, if you look at, in India, generally people talk about either we live in a bullock cart stage, right, or we are aspiring for bullet train, right. So the point is, if AI has to tell us in terms of, you know, how we learn in the future, how will we transform, so we cannot, I mean, leave out 900 plus million people who are living in villages, right. Absolutely. So the idea is not to make it very, very urbanized, you know, very, very kind of elitist idea that, you know, that.
That AI is only for urban, AI is only for industries, AI is only for commercial sector. So, obviously, this is a journey, right? So, you have to start somewhere. So, for example, I mean, the frugality what I was talking about, that we did not ask Gram Panchayat to invest anything, right? All they need to have a mobile phone, which any which way they have, right? And the idea is just to kind of record and upload. Obviously, there will be some challenges and kind of resistance also in the beginning. But, you know, once they get used to it, so, for example, today we are asking them to kind of upload your recording, right? The rest is done by system.
And system also has a provision of human in the loop so that we can go and correct it. Now, tomorrow, we see the next step what we will be doing, what we can do perhaps, right? When the next meeting happens, we can also populate the agenda from last meeting, right? So, what was discussed last time, what was committed, whether you are doing or not doing, right? And then everything goes to kind of public domain. so generally the people who live in city they know that when there is a RWA meeting nobody goes and attend but they all warfare in the whatsapp group in the village also it’s not easy to bring people but once they start getting the hang of it that okay there is a meeting I am getting the mom and it’s available in the public domain we are using AI, AI is for good AI can also be leverage for rural sector why it has to be very very elitist only for passport save so that’s just a beginning it’s just a journey and also if you see from idea point of view phenomenal idea for ministry of panchayati let me congratulate sir and the entire team to think of something like that because AI is all about idea and use case if you have the right idea you can do wonders but you have to have idea and muscles to execute it so that way I believe that this whole documentation will do wonders for them.
Gram Panchayats will also realize something which was missing in the most part of the world that you know the record keeping accountability, transparency so and so forth because generally these decisions were taken by some people only and executed by some and the large population was largely kept out of it knowingly or unknowingly right. So I think that’s what I said that you know it will change the way they work, it will change the way they think because this is only for a you know kind of we are starting only with a let’s say meeting but now they will start thinking and there will be demand from states and otherwise right what more can be done with AI.
So broader scales would be achieved. Yeah Sabasar is an example like Praman we are doing we have launched this Pancham you know bought also for all elected and selected representatives so I think it’s a great you know kind of experience efficiency would obviously help them adopt it. I mean let me tell you in our own corporate meetings we are still some of us making note. despite being on teams despite using co -pilots despite having all tools at our disposal but we are still using it we expect a junior guy to take notes and circle back so that’s a cultural change which you have to also see and these changes and these changes couldn’t have been possible if we wouldn’t have the infrastructure like Bhasni because ministry on its own how ministry got benefited we have infrastructure like Bhasni we have the GPUs got available to us through the NDIA mission otherwise procurement itself could have been a big challenge we have a team to kind of build applications so I think it takes a village to move something so that’s what has happened here
thank you for sharing your thoughts just continuing with that the department of drinking water and sanitation has actually approached us that the meetings of their village meeting VWC’s village water committees. They want to use Bhashini for that and there has been some initial interaction between the two teams.
That’s commendable, I would say. That’s awesome. So Alokji, let’s talk of some implementation challenges in rural India with AI. AI in rural governance is transformating, but complex. So what are the biggest operational challenges, infrastructure, though a bit, I think Amitji was about to share that, but then infrastructure, training, dialect diversivity and connectivity. So what challenges are you facing? How receptive are panchayat functionalities and rural citizens to AI -enabled systems?
Challenges, of course, there are many and you would have anybody tell you. What we have found out, the adoption of e -gram swaraj by our villages gram panchayat and then we have A case in point, Uttar Pradesh has got something like 59 ,000 Gram Panchayats. And for Uttar Pradesh to onboard eGram Swaraj seemed like an impossible task because it involved registering your digital signing certificates and then everybody agreeing to completely dispense with checkbooks. All your payments were then going to be, can you imagine Uttar Pradesh did it in 40 days flat. All 59 ,000 Gram Panchayats. So my point was that if you are ready with a product that addresses their needs and it is friendly and it meets, of course, my need was that I needed the money well accounted for and their need.
It was a system that could make it very easy for them to do it. So we met halfway and if UP can do it in 59 ,000, I am not prepared to hear an excuse from any other state in the country. It’s a trial by fire. Likewise for Sabasar, Sabasar again I said initially that there was a demand that was indicated from the state. So when we set out to meet that, we were clear what is it that we are looking for and people were so forthcoming. In fact, Bhashani also enabled me to write letters to the states in their languages and people were gushing with affection and what not. I got a letter in Telugu for the first time and all that.
So there are challenges but then the Ram Panchayats are predisposed to meet you halfway. So you need to begin that journey and we have seen that with regard to a number of things. There have been campaigns. Every year they carry out a campaign from 2nd October to… the 31st of December, which extends to January typically, where all two and a half lakh gram panchayats prepared their gram panchayat development plans and uploaded on the portal. So 2 .5, 250 ,000 gram panchayats, all of them planning for the next year. And so before you enter the next financial year, their plans are ready. I mean, we don’t do it in the departments, in the ministries. And all these gram panchayats have not done it once, twice.
They started in 2018. They’ve continued to do it ever since. In the COVID year, there was a request that we don’t do this campaign. So there was a massive pushback from the states that, no, we want to do it. The inertia was so great that they still did it. So there are challenges. But if we make an application like he was saying, that this is a simple recording device, this is a mobile phone, there aren’t things that you need to procure to set it up. So if you make a simple tool, people would grab it with both hands. So I think that is the embracing of challenges rather with the response we are getting with Bhashani.
So for ministries delivering last mile services such as Ministry of Rural Development and the Ministry of Agriculture and Farmers Welfare, what lessons from MOPR’s AI journey would you share? How important is open architecture and interoperability in your sense?
That is dangerous territory. I am not in a position where I could start advising anybody because they have got pretty robust systems of their own. If you look at Manrega Soft and the PM Avas Yojana, because they are running schemes which are very pointed. Avas Yojana is just about houses. Manrega is a scheme where there is of course it is as large as the things that you do in the Finance Commission. It is a very big scheme. It is a very big scheme. but it is fairly well organized and in all of these typically the beneficiary is the individual. In Panchayati Raj mode there are individuals at the end of it but our emphasis is on the institution, the Panchayat and not just E.
Gram Swaraj and the things that we do for their accounting and planning. We also hooked up with the meteorological department and there are daily forecasts being generated for every Gram Panchayat. This people are able to see on their phones and all with the similar ability as they are able to see everything using Bhashini. So it’s a great enablement all around and it can only get better.
Absolutely. So Amitji over to you. How critical is open architecture ensuring long term sustainability? And avoiding vendor lock -in.
if I can take a minute and talk about the previous question please go ahead sir rightly mentioned that different ministries have got a different mandate it’s not an apple to apple comparison but see you also have to see the panchayati raj the main role of panchayati raj what I understand is the mobilization because they are not running major schemes on their own compared to others and generally the best practices doesn’t have to be in form of technology or architecture only the idea is that if you go down from top there are two different ministries and if you go to the village you will see the same infrastructure, same set of people are only working from both departments right so the idea is if one can do others can also do So there is a lot of learning in terms of method that how we could overcome, how could we mobilize, how we could implement some of these solutions.
And I’m sure we know that RD and agriculture are also doing a lot of things, but their mandate is much bigger. But they can also, you know, take a lot of pride or kind of learning from the success which we have, right? What was the second question? The second one was that how critical is open architecture in ensuring long -term sustainability and avoiding the window lock? So you must be hearing this word called sovereignty quite a lot, right, nowadays. So the whole idea of, you know, being sovereign in any part of the, you know, technology, be it defense, be it IT, be it any way, it’s a survivability, right? So the idea is despite, in spite any kind of, you know, geopolitical risk, we should survive.
Yeah. Our system should run, right? So for that, generally, people confuse sovereignty with also making India local, et cetera. So that’s not the case, right? We will always have. some technology from outside. But we have to design in a way that it is kind of ready to shift, right? So either from a technology point of view, we have the interoperability, the standards which we have chosen, the models which we have chosen, the infrastructure which we can move around and the teams which can control, right? So the data residency has to be within India and data is with us. So obviously if we have trained on one, we can train on something else also. So the idea is also to look little bit long term.
See, what has happened that when we started, obviously there were a lot of POCs. Nobody knew, right, how AI will behave. Still we don’t know. Still we don’t know, right? I mean, so obviously that you have to start somewhere, right? And then you have to also ensure that in future, when we start with one use case, it becomes easy, right? When the department itself becomes fully AI enabled and we have 10 AI use cases running, then it becomes a problem, right? Problem of management. So that’s where I think we need to plan better for future. so that we plan. I mean, it’s not that a use case is defined. Then we found an easy method of procurement of infra or the model which I knew.
So going forward, I think there will be a platform approach. So where we have to think for future also that, okay, these AI cases are likely to come in future as well. Different kind of AI, be it agentic, be it gen -AI, be it conversational, be it computer vision analytics. And accordingly, we have to have open architecture like the way we did in a normal digital transformation. Even digital transformation, there used to be time where we created our own independent monolith applications. But now we are creating applications, you know, which are more API -based, can integrate with anybody, right? And futuristic, can scale our modular. So same concepts have to be used for AI initiatives as well.
Well said. So I think adoption comes with responsibility and that’s what you are scaling at, looking at the future. So Alokji, Sabha sir demonstrates how language AI can power grassroots governance. After Sabasa’s success, what deeper integrations do you envision with Bhashini and what does the next phase of collaboration looks like? Let’s talk about that.
And we would like through, and people are going to be speaking in any number of languages. I think the next step, my government is something that has already been very, always been very invested in providing services to making ease of living easier, as it were, and providing all manner of things. Everything is finally a service. You need to look at a doctor. You need your road fixed. You need a street light to be working. You want the log water to be drained or something. She needs more attention than us. Yeah. Okay. Over to you.
So people should come to expect. they should demand these services from their gram panchayats. There are mechanisms of doing that because gram panchayats don’t have a lot of resources in terms of manpower, in terms of people who are at their beck and call to carry out the activities that are flowing from the charter. So there are systems in a lot of these villages. You have common service centers in some states. They have their own system of common service centers like UP, like Bapuji Seva Kendra in Karnataka, like Mi Seva. So we need to take that further and we need people to be able to talk and find out if a certain service that is available to them, can they avail it in their village?
If they are to do that, what is the mechanism? And if they’ve already made an application, that what should be able to tell them that where that thing currently stands? so that is a very wide area like I said that there are a number of services we also learnt of a pilot that was carried out in Guwahati where the bus used to have a camera it used to drive through capture all number of images and basis that it would assign issue labels to them as it were if there is a drain overflowing so it takes note of that if there is a pothole then it takes note of that and then it assigns it to all these agencies whose job it becomes now to fix that so not that but maybe we have a mobile interface called Meri Panchayat which ports a lot of information from E Gram Swaraj Meri Panchayat also has the capability of capturing images of the issue that is being reported I think the next step is that image it makes sense of the image and it assigns it to the necessary department.
There are people who are mapped whose job it is to carry it out and within a certain amount of time it doesn’t happen, then there is escalation. We need to go deeper into that system. That, I think, is the next frontier. And, of course, because it involves vocalization of your demands, so bhasini is absolutely critical in this. So when we say there is a long way to go, I think that phrase is no more relevant. It’s a short way, but not even a big journey, an intelligent journey to move ahead.
So India is building public digital infrastructure for AI at scale. So how do we balance scale with accountability and public trust? We have talked much about how we are building things. But let’s talk about the other side. And can India lead the world in population scale? Of course it can. I am sure about that. But then multilingual AI for governance, when it comes, if you would like to have a shot at it first,
so one thing you all have to realize that whatever we do is a population scale and unparallel right because of our size so even our POCs exceed the kind of performance of European countries our UP sir talked about UP 60 ,000 panchayats if you look at UP maybe it will be in top 10 country right sir in terms of population and size I think the world is vouching for us when it comes to the use cases yeah see if you look at that we have got that scale now we have the experience behind us right we did Aadhaar, we did UPI we did Fastag, we did GST and we did Income Tax so now we have that confidence behind us that we can do anything of scale and with the same Prugal approach we will do 10 times cheaper than Western world and certainly not worse better only right so in terms of that and also from last decade we have evolved right so for example the concept of privacy like dpdp act consent based usage like you know adhar brought so a lot of things have improved from a policy side of it now now once you have policies in place systems are easy because system themselves act as a rule you know once you have policies in place then you don’t need so much of human intervention or discretion so since we have done it since we have kind of you know done so much so now if you look at the very simple case bhashni i remember four or five years back and i and amitabh used to i mean kind of debate also whether we need a bhashni okay right because we we had some of the google translate services so on for forth right but the idea is that i mean in the hindsight that was the right call right in future we have to have something called sovereignty word right we have we don’t have to dependent I mean we need to be frugal and we don’t want to use you know the applications which are very expensive from a taxpayer money point of view so similar things we have done a lot right so I think the next step for example if you look at roam around in AI summit you will see how many LLMs and SLMs we are building on our own right honorable ministers talked about five layers application I think we have ample talent to build applications LLMs we use open LLM but we are developing our own and Bosnia also like one of the common infrastructure energy will take care right infra and chips anyway will have dependency but that’s the rest of the world also has a dependency right not that everybody has a rare earth and everybody is building chips so that way I believe that you know that and because we have that technical know how also I mean that’s our kind of bread and bread and butter now a day right so we’ll be able to take the learnings from all these systems and we’ll move forward as of now we were a bit slow in last year or two because AI itself was new for everyone so we took some time but now I think from this year onwards we’ll really scale it up because we have tested the blood, we have seen the success and we will scale it up
sure, thank you for sharing that so as we come towards the closure of this conversation I would like to leave with one final thought which is like if Panchayati Raj institutions are the foundation of democracy can AI when built on a public stack and powered by language inclusion become the strongest enabler of participatory governance in 21st century just closing thoughts from you both Alokji, would you
absolutely he was just telling you that that we’ve been able to do things at scale this thing about UP that I told you I wear it like a badge that to have done it in some place so and it’s not an easy ask because there are so many stakeholders they’ve got various kinds of issues of their own you’ve got to engage with them address those things and if my problem is well defined and if I know what kind of a thing is going to help me redress that like Bhashini did for us I think that what you said is going to come true because that is so being able to understand my problem and knowing what parts of the problem can be fixed in what manner using the various tools that are available that is the key and it’s not an over simplification but good servant bad master so that is something that stays and it is not going to land you in the right places if you just let it go around like an animal.
But then if you know where to put it, what modules to be inserted, what has been used in the background, and so that would make you more confident. I’m not really an AI person, so I’m just speaking on the strength of what I’ve learned and the experience thus far has been outstanding, partly because we’ve had a very good partner. But other than that, I am not throwing it all open out to AI. I don’t wear T -shirts saying I love AI or something, but I have a problem and it needs fixing and I need to be able to know what aspects of AI can help me fix that in the best possible manner. And that’s my thing on this.
Yeah. So like like sir said, you know, sir is not a person. Neither am I. So if you look at, you know, that he was transparent enough to share that. No, no. So look at that way that none of us were right. Because if you’re talking about AY, I’ve been doing this, you know, digital transformation for public sector for over 20 years. Obviously, there was no AI even when there was no DPI, DPG also, you know, what we kind of retrofitted with the names. Right. So if you look at the idea of Panchayati Raj itself is a participative governance. Right. That people have to assemble in the Gram Sabha and decide on the money which they are getting, how to spend and prioritize.
Absolutely. And if AI tools like Pramana and Sabha Sar and Pancham can help that strengthen, what best you know you can expect from from a participative government, from a democratization point of view. So I think this sometimes, you know, that technology becomes secondary. And in my view, most of the time, right, the ideas have to be clear in terms of what you want to achieve. and what problem you want to solve, what scale you want to solve, what are the guardrails you have to kind of, you know, also put in place. So, for example, when we do AI, that it cannot be 100 % autonomous, right? Of course. And it cannot be 100 % human in the loop also.
Because if we have each and every transaction being, you know, approved by human in the loop, then it defeats the purpose of AI. And there is no AI, right? Then we are still living in the rule -based algorithms. Algorithms. So the idea of, you know, that AI will be that we also train, monitor, have the mechanism to take complaints, have the mechanism to perfectly, you know, kind of train it better so that we improve our accuracy. So that is how AI journey. So AI journey is slightly different from the previous digital transformation journey, which were more like a transactional systems, right? So that way, I think, if you look at currently also Sabasar, I think whatever I am hearing from people, market teams also, So it is giving great accuracy, right, in terms of translation and summarization.
And I’m sure whatever there are little bit areas to improve, it will improve on its own. So we cannot stop it, right? So once we have boarded a flight, then we can only get down at where we have to, right? So I think future is bright. And also from a MOPR experience point of view, it will also, I’m sure, energize and motivate a lot many others. I can say with my experience that if MOPR can do in rural, we can use AI tools. There is no stopping for us as a nation.
Exactly. This is truly an achievement when it comes to MOPI with the government. So you want to say anything regarding this, Alokji?
I thought of another application that works. That is something we’ve been working on, which was spatial development plans. Okay. we again engaged with a lot of panchayats that were close to the highways okay so typically if a panchayat is on a national highway close to a big city and have a population of 10 000 plus then you were eligible to participate in this program okay so there were 34 gps that we involved and we got the planning and architecture colleges to prepare spatial plans for them spatial plan would be futuristic it would zone and it would you know assign it would look into the future and see how this place was going to grow it would devise road networks or something and tell people what they would become over a period of time we had a conference with with gram panchayats around bhopal building and the people were so annoyed We don’t need a spatial plan.
Over a period of time, of course, we told them what it was going to be, but we had this epiphany that people need to be able to see what the spatial plan will help them become. And then we went into the next national conference. We had for each of these 34 spatial development plans a visualization. And we showed people that if you want to become this, you have to do this. And then there was greater enthusiasm. So the people on whom this plan is, who are going to be subjected to this plan, if I could use those words. So these people, if they’re not on board, there is no way you can carry it out. And that, I think, is wide open.
And we’ve had after that. the entire state of Andhra Pradesh has gone ahead and said that all their planning is going to be spatial plans. So that is something that is amenable to AI tools. And a final thing that I remembered that lots of times we need to convey through audio video messages. He mentioned Pancham. So Pancham is a WhatsApp -based chatbot platform which allows us to have two -way conversation with all the sarpanchas and panchayat secretaries in the country. So all these people. And so if there is messaging that needs to be conveyed, if there are videos that need to be quickly created using AI tools, that is something that would be hugely effective in getting the message across in the quickest possible way.
Thank you. Thank you so much for such endeavor. insights on the Gram Panchayat and how things are working behind. Actually, I’m sure the audience was truly, they were unknown about what’s happening around and this conversation has given a new tangent to how we look at the rural development. Thank you so much Shri Alok and thank you so much Shri Amir for sharing these thoughts on Gram Panchayat development. Thank you so much for this fireside chat. Thank you. I would like to call Ms. Deepika to please felicitate Mr. Alok.
– Shri Alok Prem Nagar- Amit Kumar Language barriers prevented rural citizens from understanding governance processes, leading to the adoption of Bhashini for multilingual support Structured documen…
EventNK Goyal, President of the CMAI Association of India, presented a series of strategies for digital empowerment, including support for language preservation and the development of language technology s…
EventWe know we have 5 .8 million professionals. For example, the Tata AI Saki Immersion Programme is empowering rural women artists to use AI as a tool for livelihood opportunity. India’s AI journey is al…
EventSabha Saar has revolutionized meeting documentation by reducing the time burden on panchayat secretaries from 65% of their workload to a simple recording and upload process, with over 115,000 meetings…
EventSabha Saar was one thing that we carried out for the convenience of the panchayats and the panchayat secretaries as opposed to E. Gram Swaraj which was our selfish motive. We wanted panchayats to plan…
Event_reportingIt’s chips, chips and computing infrastructure. The next layer above it is the cloud infrastructure, the cloud services. The layer above that is the AI models. New applications will be built on top o…
EventContinuity, institutions, and political cycles Participants generally agreed that despite shifts in leadership or political rhetoric, key institutions in the digital governance space exhibit a strong …
UpdatesTo create this plan, the government will convene an interagency AI task force comprised of National Government agencies, County Governments, higher education and private sector organization stakeholde…
Resource### Capacity Building and Onboarding
EventCapacity Building and Implementation Challenges
EventCapacity building challenges in rapid innovation
EventImplementation challenges exist between excellent policies and practical application, requiring focus on capacity building and cultural change
EventAs I said, already five lakh plus visitors have already, we were just doing the estimate, I think actual number is about six, but we are just being very conservative, everything which is measured is w…
Event_reportingThis honest acknowledgment shifted the dynamic from delegates criticizing the text to understanding the constraints the Chair faced. It transformed the discussion from adversarial to collaborative, wi…
EventThe conversation maintained an optimistic and collaborative tone throughout, with both speakers expressing enthusiasm about AI’s potential while acknowledging real challenges. The tone was forward-loo…
EventThe discussion maintained a professional, collaborative tone throughout, with speakers demonstrating expertise while acknowledging the complexity of the challenges. The tone was constructive but reali…
EventThe discussion maintained a tone of “measured optimism” throughout. It began with urgency and concern (particularly in Baroness Shields’ opening about AI engineering “simulated intimacy”), evolved int…
EventThe discussion maintained a constructive and solution-oriented tone throughout, characterized by: The tone remained consistently professional and forward-looking, with panelists building on each othe…
EventThe tone was overwhelmingly positive and celebratory, with participants expressing genuine affection for and commitment to the IGF process. The discussion was notably personal and emotional, with many…
EventChristian Daswon: Well, I think that’s exactly why the organization that we’re trying to build is focused on listening. Because identifying where there are gaps is an important thing that we need to b…
EventThe discussion maintained a collaborative and constructive tone throughout, with panelists generally agreeing on core principles while offering diverse perspectives from their respective industries. T…
EventThese key comments fundamentally shaped the discussion by establishing a realistic, pragmatic framework for AI governance that moved beyond idealistic multilateral aspirations. Wilkinson’s opening rea…
EventThe tone is consistently inspirational and collaborative throughout. The speaker maintains an optimistic, forward-looking perspective while emphasizing inclusivity and global cooperation. There is a s…
EventThe tone was consistently collaborative, optimistic, and forward-looking throughout the discussion. Speakers maintained an enthusiastic and inclusive approach, emphasizing partnership over competition…
EventAnd China using open source is actually very interesting because open source has a number of benefits and also risks. I don’t think it’s the answer to everything, but clearly it’s a way for challenger…
Event“The eGram Swaraj portal is used by all 2.5 lakh (250,000) gram panchayats and was initially built only in English, creating a language barrier for rural officials and citizens.”
The knowledge base states that eGram Swaraj encompasses all 250,000 gram panchayats and originally operated solely in English, which created significant participation barriers [S1] and [S2] and [S7].
“Bhashini, the government’s AI‑powered translation engine, allows a panchayat member to view portal pages in his own language with a single click, effectively translating directly between Indian languages.”
S1 mentions the integration of Bhashini with eGram Swaraj for translation, and S78 describes Bhashini’s capability to translate directly between Indian languages without using English as an intermediate, confirming the single‑click multilingual access claim.
“In 2019, during a Karnataka Gram Sabha, officials could not follow proceedings because they were presented in English, illustrating the language barrier.”
S2 references a 2019 event when a programme was being started and notes that the eGram Swaraj portal worked only in English, providing contextual support that language barriers were evident at that time, though it does not specify Karnataka or the exact incident.
The discussion shows strong convergence among the participants on the need for multilingual, low‑cost AI tools that are supported by capacity‑building, human oversight, and open, standards‑based architectures. These shared positions span inclusion, transparency, scalability and sustainability.
High consensus – the speakers largely agree on the principles, technologies and policy approaches required to make AI‑enabled rural governance inclusive, accountable and future‑proof, suggesting a solid foundation for coordinated action across ministries.
The discussion shows strong consensus on the need for language AI, transparency and capacity building, but reveals moderate disagreement on how much automation should be trusted without human review and on the strategy for scaling AI across ministries. These divergences are more about implementation philosophy than about end goals.
Moderate disagreement; while goals are aligned, differing views on automation safeguards and cross‑ministerial architecture could affect the speed and uniformity of AI rollout in rural governance.
The discussion was driven forward by a series of vivid, experience‑based insights that moved the conversation from identifying language barriers to showcasing concrete AI solutions and their scalability. Alok’s real‑world anecdotes and success stories (language translation, meeting minutes, solar potential, rapid onboarding in UP, spatial planning) repeatedly opened new thematic avenues, while Amit’s reflections on frugality, cultural change, and open architecture added depth and strategic perspective. These pivotal comments reframed the dialogue from a simple description of tools to a broader debate on inclusivity, sustainability, and India’s capacity to lead at population scale, ultimately shaping a narrative of confidence and forward‑looking integration.
Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.
Related event

