After a day full of collaboration and cross-learning, we were ready for another day to continue the conversations and learning.

Day 2 started with a talk by Tejas on LLM Vs LLM. With LLM being the most talked-about topic in the industry right now and NGOs showing interest in incorporating its capabilities into their chatbots, we felt the need to educate NGOs on the different LLMs available, how they differ, and provide them with a space to try them out themselves to determine which LLM is better for their use case based on different factors such as accuracy, speed, cost, etc. This session began with a presentation to share basic information. Following that, NGOs were given different prompts and use cases based on their sectors and encouraged to try their hands on https://chat.lmsys.org/ to compare different LLMs. Each organization had different takeaways based on their experience. Overall, it provided a good understanding of LLMs for both newcomers and seasoned professionals in the room. The session was followed by free work time where everyone was encouraged to collaborate with each other or the Glific team.

Before lunch, we had another session titled “Future of the Glific Platform”. We have multiple problem statements in the backlog, and we felt the Sprint would be a great time to prioritize and gather use cases from the NGOs present, as there were 15 NGOs in the room. We created a form with a list of all ideas/problem statements in the backlog and asked organizations to rank their top 3 priorities. We then had a brief discussion around the commonly selected top 3 to dive deeper into them. Overall, it gave us a good sense of a few top priorities (like the ability to resume a flow from where it stopped when another flow like help or settings is triggered; having a GPT node to easily plug in and use, etc.).

We are often approached by NGOs with different ideas/suggestions. So, after this session, we conducted another brief activity to gather more ideas from NGOs, not just on the platform but on our overall services – support, documentation, sprints, online webinars, etc. We divided the room into 3 tables of 5 NGOs each. Each table was given a pack of sticky notes and sketch pens. Whichever team came up with the maximum number of ideas within 10 minutes was the winner. Team C (KEF, TAP, Civis, Top Parent, Piramal Foundation) came up with 37 ideas and won a box full of chocolates. We also asked organizations to go around and add +1 wherever they resonated with an idea so that we could get a sense of common priorities. This activity definitely gave us a lot of ideas or problems to work upon!

After lunch, we had amazing talks by 3 NGOs – INREM, Bandhu, and Civis. All three NGOs spoke about different and interesting ways they were using LLMs. Atharva from Civis spoke about how they conduct response analysis and reporting and how they have built an in-house Average Word Embedding Classifier Model. This development helped them accurately analyze public feedback. Jacob from Bandhu shared their inspiring work on Multilingual Conversational AI integration, through which they are able to share live housing info with blue-collar workers from their database. Kiran from INREM shared their experiments in building crowdsource data using interactive location sharing, get & geotags of maps. It definitely gave a lot of ideas and inspiration to others in the room.

Even today, our friends (Rukmini & Tanushree) from the Saturday art class had organized a fun energizer which indeed energized us. We ended the day with a closing circle led again by Sheetal, where everyone shared their takeaways from the sprint.

Though there were multiple takeaways for me, I would like to share my top 2 highlights from the whole sprint:

1. Mumbai-based organizations (Bandhu, Saturday Art Class, TRRAIN) discussed that they should meet often in Mumbai to work together and learn from each other. (By the time, i wrote this blog, Jacob from Bandhu already started the WA group :))
2. While Civis & TAP were discussing sharing (export & import) some flows which are beneficial for each other, they suggested that we should facilitate this for all organizations, as they want the community to benefit instead of just two organizations.

We at Tech4Dev & Glific consistently work towards enabling collaboration and building/strengthening ecosystem. So, seeing this small yet powerful collaboration made me extremely happy.