Meet Our Award Winning AI Companion
Why AI ?
As Udhyam Shiksha works with India’s youth at scale — 3.9 million students across 12 states, and 40,000 educators so far — we face a mentorship bottleneck: teachers can’t provide timely, personalised, and domain-specific feedback, leading to incomplete projects.
To bridge this gap, we’re using AI to deliver equitable, 24/7, context-aware support—evaluating ideas, prototypes, and pitches; giving rubric-based feedback; and automating routine tasks so teachers can focus on meaningful coaching.
In short, we use AI to make high-quality mentorship accessible to every student and teacher.
A Mentor for Students, A Co-Pilot for Teachers!
What does it mean for students?
It acts as a 24/7 mentor—sending timely nudges and resources, answering project queries in multiple languages, evaluating submissions (ideas, prototypes, pitches), and giving personalised feedback. This builds a continuous build–learn–refine loop that strengthens agency and skills.
What does it do for teachers?
It’s an AI co-pilot—offering timely guidance and resources, resolving curriculum queries, automating reviews, and tracking student progress. This helps them provide the right support more effectively.
Key Features of the AI Companion
The Impact the AI Companion is Making
➜
Current user base:
Of 17 lakh students in the program this year, the AI Mentor has been rolled out to 8 lakh students and 30,000 teachers across 6 states (programs launched in Aug ’25).
Of 17 lakh students in the program this year, the AI Mentor has been rolled out to 8 lakh students and 30,000 teachers across 6 states (programs launched in Aug ’25).
➜
Active users:
3.5 lakh students and 25,500 teachers.
3.5 lakh students and 25,500 teachers.
➜
Early engagement:
~85% of teams have begun milestone submissions (vs 76% last year), marking a 9% rise, with a projected 15% increase by December-end.
~85% of teams have begun milestone submissions (vs 76% last year), marking a 9% rise, with a projected 15% increase by December-end.
➜
Curriculum adherence:
97% of teams submitted two project ideas (vs 51% last year), showing a ~100% improvement.
97% of teams submitted two project ideas (vs 51% last year), showing a ~100% improvement.
➜
Iterative learning:
~10% of teams are revising and resubmitting work based on AI Mentor feedback, even after valid initial submissions.
~10% of teams are revising and resubmitting work based on AI Mentor feedback, even after valid initial submissions.
➜
Quality improvement:
Overall submission quality is expected to improve by 50% over last year.
Overall submission quality is expected to improve by 50% over last year.
How Do We Plan To Make This Sustainable?
➜
Roll-out:
Implemented at scale across 6 states after 3 pilot programs last year.
Implemented at scale across 6 states after 3 pilot programs last year.
➜
Current focus:
Enhancing usability and expanding to all state programs.
Enhancing usability and expanding to all state programs.
➜
Development Approach:
Build the most effective solution, evaluate usability and impact, and use it for performance benchmarking and cost optimisation.
Build the most effective solution, evaluate usability and impact, and use it for performance benchmarking and cost optimisation.
➜
Tech exploration for sustainability:
Testing smaller and open-source LLMs and developing an in-house SLM for specific use-cases.
Testing smaller and open-source LLMs and developing an in-house SLM for specific use-cases.
➜
Support:
Backed by the AI for Global Development Accelerator (AI4GD) program by The Agency Fund and OpenAI.
Backed by the AI for Global Development Accelerator (AI4GD) program by The Agency Fund and OpenAI.