Artificial intelligence (AI) is quickly moving from experimentation to deployment in India’s social sector. In this direction, OpenAI has set in motion a four-city nonprofit workshop tour aimed at accelerating AI adoption among organisations that are serving rural communities. This is a tangible shift, acknowledging the gaps in AI’s promise and the gaps between pilots and scale.OpenAI’s Nonprofit Jam, which kicked off in Bengaluru on January 15, is the company’s attempt to reach out to non-profit organisations at their doorsteps. “In many ways, it’s a strategic precursor to our February AI Impact Summit. With the main theme being impact, we wanted to engage with the nonprofits more deeply. This is primarily to accelerate the integration of AI into the social sector, and we’re doing this in partnership with Karya and the Wadhwani AI,” Pragya Misra, who heads strategy and global affairs for OpenAI India, told indianexpress.com.
For Misra, the tour addresses a dissonance she has observed since joining the company in April 2024. “I have seen, whenever we have had interactions with nonprofits, it is the same set of people that show up,” Misra said. “They are really leaning into the technology, but I felt the depth was lacking on AI adoption. Our nonprofit ecosystem is very vibrant, but integrating AI into their workflows was lacking depth.”
Misra explained that the Nonprofit Jam has been designed to expand beyond the usual early adopters and reach organisations across different corners of India. This is also an acknowledgement that India’s nonprofit sector has, somewhat unexpectedly, outpaced private enterprises in embracing AI tools.
“We saw much more leaning in and adoption in the nonprofit and the social sector in India than we even saw in the private sector,” Misra noted. “That is because the social sector is always trying to get the best out of the limited resources it has.”
The deployment conundrum
The shift from pilot to deployment remains a challenge. When asked what could be the most common reason Indian nonprofits get stuck at the experimentation stage, Misra identified commercialisation pathways as the primary bottleneck.Story continues below this ad
“The pilots have an immense amount of impact, and then from there, the path to commercialisation is not very clear. When it comes to scaling and then rolling it out pan-India, or at a national level, it often becomes challenging,” she said.
Another key reason is that the nonprofit ecosystem works collaboratively. Essentially, they may have strengths in a particular sector and in a particular state. But when they want to take it to a national scale, there are often limitations in terms of the ecosystem and having partners in other parts of the country who can meet them where they are at, as far as technology is concerned.
Pragya Misra, Head of Strategy and Global Affairs for OpenAI India, at Nonprofit Jam in Bengaluru.
When asked about what AI capabilities nonprofits in India will be adopting going forward, Misra noted that the company has been witnessing some incredible examples of how its models are affecting change in the real world. The OpenAI executive shared examples of Nura Health, Rocket Learning, Agami, and Udhyam Learning.
“From Udhyam’s Saathi helping shy girls build confidence through mock pitches to Rocket Learning using AI for auto-grading and lesson planning, AI takes over the monotony so humans can focus on what matters,” Misra said.Story continues below this ad
She added that in all these processes, there is a human in the loop. The technology does not replace people; rather, it works with them, enhancing confidence, creativity, and teaching by keeping a human firmly in the loop.
What’s happening on the ground?
For the Nonprofit Jam, OpenAI has collaborated with Karya and Wadhwani AI as knowledge partners. The sessions in the series are aimed at helping nonprofits move from the experimentation stage to deployment. Participants get to work with experts from OpenAI to understand how tools like ChatGPT can support programme delivery and improve decision-making.
The gap between ambition and practicality runs deeper, according to Chintan Donda, a senior machine learning engineer at Wadhwani AI, which serves as a knowledge partner for the initiative.
“Nonprofits are increasingly setting ambitious goals for using AI, such as predicting crop failure, identifying students at risk of dropping out, or enabling automated health triage,” Donda noted. But translating these ideas requires navigating challenges unique to the social sector.Story continues below this ad
According to Donda, despite India’s digital public infrastructure (including Aadhaar, UPI, and language platforms like Bhashini), much ground-truth data is still collected manually through physical registers or basic spreadsheets. The Digital Personal Data Protection Act has introduced privacy and consent requirements that many nonprofits are still building capacity to handle, particularly when working with marginalised communities.
Moreover, talent limitations push organisations toward general-purpose models like ChatGPT or Gemini. “While these tools provide quick access to AI capabilities, they may not always offer the level of localisation or customisation desired for specific social-sector contexts,” Donda explained.
Designing for India’s realities
Donda asserted that effective AI deployment in India’s social sector needs the shedding of imported assumptions. “There is a need to shift the focus from sophisticated models to building resilient AI systems that work in real-world conditions,” he said.
The human-in-the-loop model proves critical, and voice-first design matters more than sophisticated interfaces. “A high-end web dashboard is often unusable for field workers operating in remote areas with unreliable internet access,” he noted. Deployment through WhatsApp or IVR systems, in local dialects, ensures adoption by fitting into existing workflows.Story continues below this ad
On the other hand, when asked how OpenAI measures the success of nonprofits, Misra shared that it will be measured not by workshop attendance but on whether organisations can serve more communities months from now. “The real value of this technology is going to be unlocked in tier-three rural communities, and that is where we are seeing a lot of impact coming in,” she said.
The metrics remain qualitative: Can organisations serving lakhs of people double that reach? Can they add new verticals because AI handles existing work well enough?
“It is hard to put a number to these things,” Misra acknowledged. “But because we stay close to the nonprofits themselves, they often come back to us and share these heartwarming stories about how our models or the work that we are doing collaboratively is really changing and helping people on the ground.”