1. How long have the founders known one another, and how did you meet? Have any of the founders not met in person? We have known each other since 2015. We met during our undergraduate degree in Georgia Tech and were also roommates. 2. Who writes code, or does other technical work on your product? Was any of it done by a non-founder? Please explain. The code and technical work for the product is done by both of the co-founders. Vignesh handles the web application while Likhit is responsible for the AI models. 3. Please record a one-minute video introducing the founder(s) 4. Company name Ira Project 5. Describe what your company does in 50 characters or less Create scaffolded AI-proof learning activities 6. Company url iraproject.com 7. If you have a demo, attach it below 8. Please provide a link to the product, if relevant. app.iraproject.com [email protected] | asdf1234 9. What is your company going to make? Please describe your product and what it does or will do? We are building a learning platform that empowers students to build skills beyond AI capabilities. Ira Project does this in a two-pronged approach: 1) Leveraging AI to scaffold learning for skills that AI excels at. We do this by using Ira as a thinking partner that learners have to teach concepts to. 2) Designing assessments that the state-of-the-art LLMs perform poorly on. We do this by generating a set of incorrect answers and ask learners to identify the exact line of reasoning that would produce each of those incorrect answers. 11. Where do you live now, and where would the company be based after YC? Bangalore, India / San Francisco, USA 12. Explain your decision regarding location. The founders are currently based out of different locations. Vignesh lives in Bangalore, India, while Likhit is in Atlanta. We want the company to be in San Francisco so we can be close to the cutting edge in AI. 13. How far along are you? We have a fully classroom ready product for Physics and Math including fine-tuned AI models. We have completed a pilot at an IB school in South Korea and have five more pilots in progress across the globe. 14. How long have each of you been working on this? How much of that has been full-time? Please explain. Vignesh has been working on the product full time since April. Likhit has been working full time since May. 15. What tech stack are you using, or planning to use, to build this product? We're using Next.js deployed on Vercel for our web application. For Ira, we have built out our proprietary symbolic computation engine that integrates with a fine-tuned LLM. 16. How many active users or customers do you have? How many are paying? Who is paying you the most, and how much do they pay you? We have one school (around fifteen students) that is using the product as a pilot. There are no paying customers yet. 17. If you are applying with the same idea as a previous batch, did anything change? If you applied with a different idea, why did you pivot, and what did you learn from the last idea? - We have rolled out a new assignment type that AI models (GPT 4-o, Sonnet) struggle with - We have built out assignments in Physics and added capabilities for image generation and graphs - We have identified that IB schools are good early adopters - We have fine-tuned and improved our AI models to better scaffold learning for students based on pilot data 18. If you have already participated or committed to participate in an incubator, "accelerator" or "pre-accelerator" program, please tell us about it. No accelerator with Ira Project. 19. Why did you pick this idea to work on? Do you have domain expertise in this area? How do you know people need what you're making? The idea for Ira Project came out of Vignesh's needs while running his non-profit. He was worried that students weren't learning 21st century skills like critical thinking, problem-solving and conceptual understanding. Learning today focuses on solving procedural tests, which AI excels at. We validated this problem statement by talking to educators in our network. We then ran several experiments consisting of more than 1000 emails and LinkedIn messages to teachers and school administrators. We also demoed our product with schools around the world and completed one pilot. In all our interactions, we noticed that there is a need to prepare students for an AI driven future, and schools are actively looking for a solution. 20. Who are your competitors? What do you understand about your business that they don't? Kognity, Amplify, Khan Academy There are two reasons why we think Ira is a better solution to prepare students for an AI future - 1. Current solutions are replicating the existing "banking model" of education where technology and AI is used to deposit or put knowledge in children. Instead of letting AI "program" students, we think it's important to have students "program" AI. 2. Most platforms penalise students for mistakes and aren't built for productive failure. Our platform encourages risk taking and ensures students learn from their mistakes immediately. 21. How do, or will you make money? How much could you make? Currently, we are planning to charge schools to use our platform. For each course we plan to charge $500 per year. 22. If you have not formed the company yet, describe the planned equity ownership breakdown among the founders, employees and any other proposed stockholders. If there are multiple founders, be sure to give the proposed equity ownership of each founder and founder title (e.g. CEO). (This question is as much for you as us.) Incorporated a Delaware C Corp Vignesh Prasad, Chief Executive Officer - 43% Likhit Nayak, Chief Technology Officer - 43% ESOP Pool - 14% 23. If you had any other ideas you considered applying with, please list them. One may be something we've been waiting for. Often when we fund people, it's to do something they list here and not in the main application. By leveraging the data from activities like having students teach AI, we had a few ideas: 1) Offer a SaaS tool which accurately predicts exactly which questions students might get wrong and consequently, predict their perfomance in exams. It then uses this to automatically generate questions for each student. 2) A personal tutor trained on language data from each student that provides Socratic dialogue 3) While there exist a lot of benchmarks that assess the performance of LLMs on their ability to correctly answer questions, there is no benchmark to assess their capability as a teacher. This could serve as a new benchmark for AI tutors of the future 24. What convinced you to apply to Y Combinator? Did someone encourage you to apply? Have you been to any YC events? Our eventual goal with the Ira Project is to change the education system at scale. This would require both systemic and institutional change and as a start-up with this ambition, we would need to persevere despite many potential failures/challenges. YC's expertise, their hands-on approach, and the no-strings-attached funding model, would give us the flexibility and resources needed to experiment and pivot as we fine-tune our product to eventually drive this change. 25. How did you hear about Y Combinator? We initially heard about YC from our friends in college.