1. What problem are you solving? Generative AI has fundamentally changed the way the world works and has made current assessments redundant - Traditional assessments are multiple choice questions which look for procedural knowledge that will become irrelevant in an AI future. - Learners can use AI to bypass these assessments easily 2. What product are you building to solve this problem? We are building AI proof assessments that test for 21st century skills. Rather than asking a test taker to answer questions our assessments require them to explain concepts to Ira, our AI. Using the explanation given by them, Ira attempts to solve a set of questions. The test takers iterate on their explanations till Ira is able to solve all the questions correctly. While teaching concepts students have to exercise critical thinking, metacognition and communication skills. 3. What unique insight do you have into this problem? Why are you choosing to focus on it? The core insight behind Ira Project comes from the Feynman technique - if you cannot explain a concept simply, you do not fully understand it. Rather than using AI to teach students we want students to teach AI. We chose to focus on this problem when Vignesh experienced it while running his non-profit. As his non-profit grew to 10 learning spaces and 300 children he found that he could not assess his students effectively. 4. How much traction do you have, in terms of revenue and number of users? We have no revenue so far. We have piloted the product in schools and after school spaces in India and have tested with 50 users. 5. Who are your competitors? How do you differentiate from them? Other companies in the assessment space include ETS, Quizizz, Pearson, etc. Our assessments are better as compared to our competitors for four reasons - 1. Language data gives a richer insight into a student’s thinking as compared to the numerical data or multiple choice data of traditional assignments. 2. LLMs have not been trained on our new type of assessment allowing us to detect AI generated responses. 3. Since we give student’s answers and do not penalise them for mistakes our assessments disincentivise cheating. 4. Students actually learn as they take our assessment. 6. How big is the market opportunity? Provide a bottoms-up calculation of how you estimated the market size. There are four areas where assessments are used - - K-12 Education: The average schools spends $150 per student per year on digital tools. Assume 10% of these tools are assessment platforms. Average school size in the US is about 500. There are 100,000 schools in US alone. Market Size = $150 x 10% x 500 x 100,000 ~ $1 Billion. - Higher Education: There are about 20 million undergraduate students in the and each student spends $20-$80. Taking an average of $50 per student we can calculate Market Size = $50 x 20 million ~ $1 Billion - Standardised Testing: SAT and ACT are the big testing tools in the US. Approximately 2.5 million students take these tests each. The tests cost $150 approximately Market Size = $150 x 2.5 million ~ $375 Million - Pre Employment Testing: Since this is an emerging market it is hard to conduct a bottoms up calculation here. The market size here is approximately $1.83 billion as of 2023 Total Market Size in the US ~ $4 Billion 7. How long have you been working on this company? Vignesh has been working on the product full time since April. Likhit has been working full time since May. 8. Are any of the founders technical? If not, who is building the product? 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.