LivelihoodsWorkforce Enablement

Udhyam Learning Foundation: Udhyam Saathi – AI Mentor

Strengthening Entrepreneurial Mindsets at Scale through AI-Enabled Mentorship

AI Mentor
AI Mentor
rolled out to 8 lakh students and 30,000 teachers across 6 states
3.5 lakh
3.5 lakh
active student users and 25,500 active teachers
Video Thumbnail

Watch the story in 60 seconds (AI-generated video; visuals are illustrative)

Case at a Glance

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.

Read the 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
Udhyam Challenge
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
Udhyam Saathi AI Mentor Solution

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

Tools/ Techniques Used For What It Enabled Category
Glific Student and teacher interactions on the WhatsApp platform 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
Google Cloud Platform API-driven backend services using Python Scalable AI service deployment Commercial
Google Big Query Primary data store Analytics and impact measurement Commercial
Langfuse (RAGAS Evaluation) Observability Quality monitoring and evaluation Open Source
Google 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:

01
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.

02
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.

03
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 Adaptation

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

See it in Action

Solution Video
Solution Video Demo

Udhyam’s AI mentor delivers entrepreneurial learning via WhatsApp, offering multilingual guidance, onboarding, nudges, and personalized feedback at scale for students and teachers.

Solution Video
User Testimony 1

A lecturer from Patiala shares how the AI Mentor chatbot has been a standout digital assistant — quick, responsive, and effective at clearing doubts for both teachers and students.

Solution Video
User Testimony 2

A teacher reflects on using the AI Mentor chatbot to support students through their projects — describing how it helps with registration, addresses common challenges, and keeps students motivated and on track

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.

  • © 2026 India Leaders for Social Sector. All Rights Reserved.