Strengthening Entrepreneurial Mindsets at Scale through AI-Enabled Mentorship
Case at a Glance
Impact
AI Mentor
rolled out to 8 lakh students and 30,000 teachers across 6 states
3.5 lakh
active student users and 25,500 active teachers
About the organisation
Udhyam Learning Foundation envisions a caring world where people fearlessly pursue their potential. It addresses India’s opportunity crisis by building agency and entrepreneurial mindsets at scale through a four-year Entrepreneurial Mindset Curriculum (EMC), implemented with state governments for students and teachers in Grades 9–12 across government schools.
Problem Statement
Udhyam’s Entrepreneurial Mindset Curriculum reaches millions of students across diverse contexts. At scale, however, teachers face a mentorship bottleneck and are unable to provide timely, personalised, and domain-specific feedback on student projects. This results in incomplete work, limited iteration, and uneven learning outcomes, constraining the effectiveness of experiential, project-based learning.
Solution
Udhyam Saathi is an AI-powered mentor designed to deliver equitable, continuous support to students and teachers at scale. For students, it functions as a 24/7 mentor answering project queries in multiple languages, evaluating ideas, prototypes, and pitches, and providing personalised, rubric-based feedback that supports iterative learning. For teachers, it serves as an AI co-pilot, automating reviews, resolving curriculum queries, and tracking student progress to enable more focused, high-quality coaching.
Quick Facts
Organisation Name
Udhyam Learning Foundation
Organisation Website
Visit Site
Founding Year
2017
Number of Beneficiaries served
17 lakh+ students and 30,000 teachers
Geography Served
12 Indian States (active AI rollout in 6 states)
Focus Area
Capacity Development Programmatic Impact
Functions Impacted
Program Delivery / Beneficiary Services Training & Capacity Technology & Data Management
Service Provider
Consuma.ai
Contact Email
contact@udhyam.org
SDG Addressed
Full Case Study
Challenge
Scaling high-quality mentorship in experiential learning programs
Udhyam’s Entrepreneurial Mindset Curriculum emphasises learning by doing, requiring regular feedback on ideas, prototypes, and projects. As the program scaled to 3.9 million students across 12 states, teachers faced increasing demands on time and expertise.
Key challenges included:
- Limited teacher capacity to provide timely, personalised, and domain-specific mentorship at scale
- Delays in feedback leading to incomplete projects and reduced iteration
- High administrative burden on teachers, limiting time for meaningful student engagement
- Inequitable access to quality mentorship across geographies and school contexts
Solution
An AI mentor and co-pilot embedded into daily teaching and learning workflows
Udhyam Saathi was designed as a multilingual, multimodal AI mentor delivered primarily through WhatsApp, complemented by a Progressive Web App for additional support. The solution enables continuous student–mentor interaction while integrating seamlessly into existing program structures
Key features of the solution include:
- 24/7 AI mentorship for students, including query resolution and personalised feedback
- Evaluation of submissions across text, image, and video formats
- Multilingual support across five languages
- Rubric-based assessment and feedback for ideas, prototypes, and pitches
- AI co-pilot support for teachers, including curriculum guidance and automated reviews
- Dashboards for tracking student progress and program impact
To ensure safe, reliable, and scalable use, Udhyam embedded strong implementation and governance safeguards within the solution:
- Rollout approach: Implemented with state governments through formal MoUs, localised for context, supported by trained teachers/local teams, and tracked via dashboards, observations, and surveys.
- Guardrails & validation: Queries are pre-processed (relevance/intent checks); irrelevant queries are declined, and unclear ones are guided with options. Rubric-based evaluation and periodic checks help maintain data quality and reliability; prompt guardrails include a Negative Test Suite for handling out-of-scope cases.
- Privacy & access control: Student PII is not shown on dashboards and is not exposed to LLMs; data is masked by default, and access is authorisation-based (approval + NDA where needed). APIs/data transfers are SSL-secured, and the pipeline is designed to be self-contained.
- Quality & risk monitoring: Performance is monitored through CSAT and observability tooling (Langfuse), along with regular audits/feedback loops to reduce bias and improve response reliability at scale.
Outcomes & Impact
Improving engagement, adherence, and learning quality at scale
- Expanded access to mentorship 8 lakh students and 30,000 teachers supported across 6 states.
- Increased engagement 85% of teams submitted milestones, a 9% increase over the previous year.
- Stronger curriculum adherence 97% of teams submitted two project ideas, compared to 51% last year.
- Improved iteration ~10% of teams revised submissions based on AI feedback
- Enhanced learning quality Overall project quality expected to improve by ~50% year-on-year.
- Improved quality of student submissions Compared to the previous cohort, student submissions showed a 59.5% increase in ideas per team, a 163% increase in novelty of ideas, and a 126% improvement in articulation of ideas
Technology Stack
| Name of the Tool | Where it was used | What it enabled | Category |
|---|---|---|---|
| WhatsApp orchestration via Glific | Student and teacher interactions | Conversational AI delivery at scale | Commercial |
| Progressive Web App (PWA) | Extended user support | Rich interaction beyond chat | Custom-built |
| RAG Knowledge System | Content retrieval | Context-aware responses using vector embedding | Custom-built |
| MongoDB (Vector Embeddings) | Knowledge base | Retrieval and logging of learning data | Open-source |
| GCP Cloud Run APIs (Python) | Backend services | Scalable AI service deployment | Commercial |
| BigQuery | Primary data store | Analytics and impact measurement | Commercial |
| Langfuse (RAGAS evaluation) | Observability | Quality monitoring and evaluation | Open-source |
| Looker Studio | Dashboards | Visualisation and data analysis | Commercial |
Key Project Learnings
Udhyam’s deployment illustrates how AI-enabled mentorship can strengthen experiential learning outcomes at scale:
- Extending Mentorship Unlocks Engagement: Introducing AI-based mentoring addressed teacher capacity constraints and ensured students received timely, personalised feedback. This led to higher participation in milestone submissions and more consistent progression through the curriculum.
- Meeting Users Where They Are Drives Adoption: Delivering mentorship through familiar platforms such as WhatsApp reduced friction for students and teachers. This ease of access enabled sustained usage and rapid scaling across states.
- Feedback Enables Iterative Learning: Rubric-based, timely feedback encouraged students to revise and improve their work. This directly contributed to higher curriculum adherence, increased resubmissions, and measurable improvements in overall project quality.
Potential for Wider Adaption
| Sector | Adaptability of the Solution |
|---|---|
| State Education Systems | Scalable support for experiential and project-based learning programs |
| Skilling and Employability Programs | Continuous mentorship for youth-led projects and skill development |
| Teacher Capacity Building | AI co-pilots to improve instructional and mentoring capacity |
Additional Details
The information provided here is created as a community resource and is not intended as professional advice or a recommendation by ILSS or Koita Foundation. While we strive to ensure the accuracy of the content, we do not take responsibility for any errors or omissions. Users should use their own discretion before making any decisions based on this information. ILSS or Koita Foundation assume no liability for any actions taken based on the information provided.