This is a first for us, starting a second cohort just a couple of months after the first cohort ended. You can read more on what we achieved during the first cohort in our AI Cohort Program 1.0 | Final Report. It was definitely our best cohort execution, and there were a few reasons why I think it went so well.
- Thanks to grants from the Laidir Foundation, Koita Foundation and CIFF we were able to give each NGO a sizable grant with the main focus of hiring a tech resource or software partner to help them move fast on AI.
- We had dedicated mentors, two per NGO for the duration of the cohort, who were savvy with AI, tech and the bigger picture. They interacted with the NGOs on multiple dimensions, including choosing the right program which could be improved with AI. The mentors spent on an average 4-8 hours with each NGO. Thanks to the funding, we could give the mentors a reasonable stipend.
- Thanks to grants from OpenAI and Anthropic, we were able to give the NGOs enough credits to run a pilot and beyond. Very recently we were also able to give NGOs a free 6 month license to Claude Pro (the coding co-pilot for developers)
- We found a common pattern in using AI for assessing student work from this cohort, and started work on the AI assessment platform with 3 NGOs in the education sector.
Given the above, we decided to keep the momentum going and started planning and fundraising for AI Cohort 2.0. We decided to continue working with 3 NGOs (Avanti Fellows, Inqui-Lab and Avni) and onboard a few more NGOs. We also decided to group NGOs into 3 specific threads to enable improved peer learning. and potentially building shared infrastructure across these themes.
- With Inqui-Lab, our goal will be to design and implement the AI Assessment platform and roll it out in a couple of districts for the next academic year, i.e. June 2026
- We have a group of NGOs focussed on Mentoring/Coaching (Reap Benefit, Madhi, U&I Trust, Avanti Fellows). We hope to learn from deployments done by Career Village Coach and Tabiya when building this solution.
- We have platforms that can benefit from adding AI to make it easier on our NGO clients. The team at Avni, did an amazing job in the last cohort of adding AI features, and we figured that Glific could benefit from this work and potentially generalize it. The Avni AI Cohort Program Report is worth a read.
In addition to the specific focus areas, we are also improving the program with a few changes, specifically working with the NGOs to bring in resources sooner rather than later/never. In the last cohort we realized that this was a significant barrier to some of the NGOs. We are prioritising text-to-speech and speech-to-text features because voice is critical for scaling access in rural India. This will help many NGOs support users across diverse languages, literacy levels, and demographics. We also decided to skip experimenting with machine learning and big data this time around, since it did not seem very replicable. In addition to the returning mentors, we’ve added a few new mentors (Poorva M, Jerome and Vinay) who bring in significant experience and skills. For this cohort, we will also experiment with multiple LLM models (Claude, OpenAI and Gemini), thanks to potential generous grants from the respective companies.
The one issue we are still working on is, how do we keep the momentum going after the cohort ends. There seems to be a significant slowdown when the cohort ends.
We are having our first virtual meeting next week, and an in-person meeting in Rishikesh at the end of the month. Quite excited to see how this cohort shapes up. The list of NGOs in the cohort are
One thing that’s new about the approach to AI cohort is that we actively look for common themes and recurring problems across NGOs, not just individual use cases. Instead of building one-off solutions for a single organisation, we ask: can this be turned into a shared solution that the whole ecosystem can use? This gives the program a strong ecosystem lens, and we hope to keep turning cohort learnings into reusable infrastructure for many NGOs, beyond the cohort.
