Rocket Learning

Over 85% of brain development occurs before the age of eight, when low income Indian children have almost no cognitive stimulation through institutions or at home. They are already falling behind their higher income peers - 43% can’t recognize alphabets, and 35% can’t recognize numbers 1-9 in grade 1. 

Rocket Learning' builds early childhood and foundational learning at scale by connecting the government system and parents, and driving community change by systemically leveraging technology, media and social incentives. We help 20,000 teachers reach over 200,000 parents everyday with contextualized content in the local language. We then invite parents to send us back their responses so that we can build continuous motivation through real-time nudges.

Our solution has demonstrated learning gains, and is inherently cost-effective and scalable - our variable costs are 10c/child/year. Globally, it can transform lives by catalyzing high quality, inclusive early learning for over 300 million young children.

Global research has established the importance of Early Childhood Education (ECE). Over 85% of brain development occurs before the age of eight. Quality ECE ensures school readiness, substantially increases school completion and lifelong earnings, and frees up parents (especially women) to enter formal employment.

However, over 35 million low income children in India can't access quality ECE. Public schooling in India and at the Pre-K level is limited to daycare centres that focus on nutrition/ healthcare rather than cognitive stimulation. Parental engagement is critical for inclusive education, but low-income parents often need supporting in building what we call "AIM" - Awareness to be involved in their children's learning at the foundational level, Information that is contextual and usable, and the Motivation to be engaged at the sustained levels necessary for adequate cognitive stimulation. 

This shows up in dismal educational statistics - 43% of low income children can’t recognize alphabets and 35% numbers 1-9 in grade 1. Delay in building neuronal connections in early years becomes more and more difficult to change through remedial interventions later in life. 

Globally, 250 million children (43%) younger than 5 years in low and middle-income countries are at risk of not reaching their developmental potential.

We partner with government systems to create online groups of teachers and low income parents for each school, into which the Rocket Learning "bot" is added. Everyday on these groups, we send parents contextualized content in the local language that they can do with their children. This content is designed with time and resource-constrained low-parents’ needs in mind - activities take less than 20 minutes and involve household materials. We then invite parents to send us back their responses on the group to create peer accountability and social "pull". We also use real-time data and technology to build continuous motivation through virtual ‘report cards’ based on how often parents participate, social media challenges, and inspiration from community and teacher influencers. 

Since we receive thousands of responses from parents everyday, we are also uniquely positioned to engage in learning engineering. We rigorously capture and analyze data on how learners and their families engage with our content. Through A/B testing, we can answer questions like - what content domains do parents respond to, what time do they respond, what length and resolution of content is most accessible, etc. This is used to continuously improve our programme.

Low income parents in India invest in and care deeply about education; however in many cases, they may not have "AIM" in children's early years- Awareness, aspiration, and confidence to be involved in their children's learning while they are in the developmentally crucial years before age eight, Information that is contextual and usable, and the Motivation to be engaged in a sustained manner. Edtech apps, including free ones, do exist but they are primarily for high-income users - parents need to download new app taking up large memory, data and graphics capacity; are all in English and use non-contextual content (eg. pizza for fractions which many children haven't seen); they also run one on one with the user and cut all moderators or community members from the interaction.

Our product meets users where they are (Whatsapp which they have already downloaded and know to use), uses local language, contextual content that we have developed in-house; runs on groups so that peer effects come into play (inspired by self-help groups) and where the early adopter parents/children inspire others around them (essential for our communities where issue is motivation and confidence). The presence of teacher moderators also helps to motivate parents since "school" is considered essential as opposed to "on-the-side learning". Thus working with a combination of government "push" and social "pull" allows us to engage a high percentage of parents and children enrolled in  public schools who need our product the most - close to 30% of parents on our platform engage regularly even one year into the programme. 

We engage with users constantly to iterate on all our design choices around nudges, social incentives, and accessibility of our content and technology. Before we launched, we conducted over 50 parent and teacher interviews which helped nuance the understanding of parent communities that we had built in our time at SEWA and Pratham. Today, all 37 members of our team conduct parent interviews once a month to get direct feedback and help us build new features (last month we conducted over 250 interviews).

Global research on early childhood education by JPAL, Caroline Hoxby, and others shows that caregiver involvement is crucial to increase the engagement of young learners. Our solution is a scalable, cost-effective way to provide communities of parents with the daily content, nudges, and social motivation to support their young children in learning on a sustained basis. We design with the needs of low-income parents in mind, which shows in our results - two-thirds of all enrolled parents are activated in our programme, and half of activated parents engage multiple times a week (very high compared to most edtech products).

Currently we work in 4 states of India with over 20,000 schools and 200,000 parents, of which over 100,000 parents (and growing) are activated, and over 50,000 are engaging with our educational content on a monthly basis. We already have partnerships in place to work with 1 million children this year through our platform. 

Our byte-sized educational content, contextual and in the local language, is currently being sent to over 5 million learners through state governments and has had over 10 million views on YouTube.

We believe this model is poised for future growth in India as well as globally, since it is an impactful, cost-effective and scalable model that uses no physical infrastructure. However we do require support on creating a more robust case for impact, advocacy, and use of advanced analytics and technology such as AI for processing children's responses.

Instead of building a new app and trying to get parents and teachers with <4GB of phone space to download and check it everyday, we build on top of Whatsapp and other platforms that they check everyday and limit our content to <5MB per video so we don't hit against their data limits.

We work with a low tech front end but a very sophisticated back end - our tech platform lets us send academic content to parents in thousands of groups everyday at the same time, rather than going through the leaky information cascade (state->district->headteacher->teacher->parent) that means most parents don't get any content at all.

We design for our parents - we send academic content that's audio visual, short and TikTok style, so that even low literacy parents can use it, contextual so it doesn't use materials or words that low income parents don't know (think: dough instead of clay, turmeric instead of yellow paint), and requires <20 minutes since we know our parents are busy.

We don't just send information but also nudges - our tech lets us send real-time "report" cards based on participation, badges and certificates, and weekly video compilations of children's activities to keep parents and teachers engaged. We're doing A/B tests and building a playbook on how behavioural change for parent communities is possible.

We hope that these innovations will allow public schools in developing markets to create a home-school digital connection in the form of a first of its kind National Parent Platform.

- Currently: 200,000 users (although our content reaches over 5 million children, we are only considered those children whose learning we impact directly and in a sustained manner)

- In 1 year: 1 million users (partnerships already in place with state governments)

- In 5 years: 50 million users

The output and outcome indicators that we are using and will continue to use to measure progress and impact are:-

1. User engagement and response rates in the parent whatsapp groups

2. Children learning outcome assessments for 5 domains – cognitive skills, physical skills, social-emotional skills, literacy and numeracy

3. Change in parental knowledge, attitude and confidence - through responses on parent surveys

Over the long-term, we would like to track impact of parental involvement on learning and attendance within schools (through reduction in drop rates, primary school completion rates and so on)

Full time team members: 37

Part time team members: 3

Our primary value is impact first – to put our communities in the centre of everything we do through – 

1. Staffing/ leadership – Currently 50% of our team are women, and 30% team members have themselves studied in government schools and come from low-income backgrounds. Senior field team leaders come from the communities we work in, and we hope to increase staffing from our communities as we move forward. We also plan to add community members to the board.

2. Human-centred design – We design our intervention not viewing our parents from a deficiency lens, but from a strengths-based lens, of best working with their strengths. We know that our parents care deeply about their children’s education. So we design to best use the 20 minutes a busy parent has, on a platform they already use and are comfortable with, sending content that they can watch and use even if not comfortable with text, using materials like bowls or beans that they have in their house. We share back positive reinforcement and strengths-based feedback, pointing out positive examples of parenting from what they and their communities share.

3. Constant feedback mechanisms – Once every month, every member of our team, from tech to media to data analysis, speaks to a set of parents and teachers to get their thoughts on the programme, its strengths and weaknesses. We then capture their feedback and discuss it as a team, using it to guide future program elements, like quizzes and media campaigns.

We are committed to diversity in our board (https://www.rocketlearning.org...) and leadership team (https://www.rocketlearning.org...). Currently, 2 of 5 board members, and 3 out of 7 leadership team members are female - our aim is to reach at leat 50% female representation in the next year in both. We represent all 5 geographic corners of India.

There are diverse skills represented in the team- Namya Mahajan led SEWA’s cooperative federation (India’s largest women’s organization), serving over 300,000 informal women workers. Vishal Sunil was co-founder and CTO of Kestrel, a Boston-based ag-tech company. Vibha Iyer was a Teach for India leader, and Azeez Gupta and Utsav Kheria ran large social impact organisations.

Diversity is one of our 7 core values as an organisation. We word this as "Ensure multiple voices in the room, including different genders, caste, regions. Especially value voices from the people closest to the field - and keep community voices at the centre ." We have taken hard decisions in order to follow this value- we had one of our senior state team leaders to undergo counseling after he was found to be making insensitive remarks to women colleagues, and asked him to leave (even at a critical work juncture) when his behavior did not improve.

Rocket Learning has proactively created policies for prevention of sexual harassment, and has mandated 3 month paternal leave, in addition to 6 months maternity leave. The paternity leave policy is based on our organisation's choice to ensure equal work and home environments.

For us, the most attractive attraction of Solve is the community of experts, mentors and peers. We are tackling a very difficult problem - that of creating behavioral change in low-income users. This will require us to innovate continuously in tech, product and user insights, and we believe that MIT and Solve will push our thinking on this through the network of experts. We are sure that we will also have to iterate and pivot as an organisation on our strategy and plans - peers doing similarly audacious things would be a great sounding board. 

These points also correspond to the first and second barriers mentioned earlier in the application. We need to work on improving the political economy pull of our product for politicians and the behavioral aspects for parents - something that will require multi-disciplinary thinking that Solve is known for. Secondly, improving the product and technology to better personalise learning while maintaining peer effects is crucial - we believe that Solve will be a great advisor and their networks will help us hire talent or partner with existing tech experts and CS professors. 

1. Human capital - we would like mentors and support in sourcing strong tech and product talent. While we have a strong 7 member tech team led by an excellent CTO, we need to get on board additional talent that is really excited about the problem at hand.

2. M&E - next to no research has been done related to early childhood care and education in developing country contexts. We are keen to leverage Solve's and MIT networks to create knowledge and insights on this crucial topic. Our board member Iqbal Dhaliwal, who is ED of MIT-JPAL, has been a great support already in this, and we need to build up a pipeline of partner experts.

3. Technology - Some of the technology we are aiming to build is really bleeding edge, and we will need partnerships with industry and academia to get there. 

Solve partners and mentors have expertise in education, startups and technology - we would like to partner on each of these aspects. The clear names that come to mind on education include the Gates Foundation and Save the Children - on funding, expertise and research. 

Dream MIT faculty advisors would include Prof. Anant Agarwal and Prof. Anantha Chandrakasan, who have strong grounding in emerging markets, education and of course technology. During my time in Harvard, I also interacted with faculty and students at the MIT Media lab, and education focused labs at MIT, and we are keen to partner in some capacity with these.

We are working to build inclusive classrooms where parents and teachers can come together to ensure the best learning for every child. By using human-centred technology in our design - for instance, using Whatsapp instead of forcing parents to download an additional app, sending local language and contextual content (for instance, dough instead of clay, potatoes instead of blocks as learning and teaching aids), and using short-form content to reduce data and time constraints for busy parents, we make it possible for busy, stressed low income parents to engage with their child's learning in a joyful and effective way. 

We receive huge amounts of data from our engaged users - more than 5 million views and 4 million responses in the last year alone. In the coming 12 months, we anticipate receiving 20 million text, audio and video responses. We would like to use the AI for humanity prize to build our capabilities to leverage this data and build cutting edge technology that will drive up user engagement and program effectiveness. 

We aim to develop advanced neural networks to come up with automatic engagement nudges and advice to parents, teachers and classroom groups based on their specific request – this will be critical as we scale to 1 million users in 50,000 schools next year. Additionally, we plan to use computer vision to sort, classify and verify image and video responses to understand content effectiveness, student learning levels and change our model accordingly while giving usable advise to parents and teachers.

The AI for Humanity funding will allow us to hire great talent and the networks will allow us to partner with leading researchers in the field who can guide us on how to use our data gold to train ML models.

We receive huge amounts of data from our engaged users - more than 5 million views and 4 million responses in the last year alone. In the coming 12 months, we anticipate receiving 20 million text, audio and video responses. We would like to use the AI for humanity prize to build our capabilities to leverage this data and build cutting edge technology that will drive up user engagement and program effectiveness. 

We aim to develop advanced neural networks to come up with automatic engagement nudges and advice to parents, teachers and classroom groups based on their specific request – this will be critical as we scale to 1 million users in 50,000 schools next year. Additionally, we plan to use computer vision to sort, classify and verify image and video responses to understand content effectiveness, student learning levels and change our model accordingly while giving usable advise to parents and teachers.

The AI for Humanity funding will allow us to hire great talent and the networks will allow us to partner with leading researchers in the field who can guide us on how to use our data gold to train ML models.

We aim to develop advanced neural networks to come up with automatic engagement nudges and advice to parents, teachers and classroom groups based on their specific request – this will be critical as we scale to 1 million users in 50,000 schools next year. Additionally, we plan to use computer vision to sort, classify and verify image and video responses to understand content effectiveness, student learning levels and change our model accordingly while giving usable advise to parents and teachers.

The GST Prize funding will allow us to hire great talent for powering the technology behind our solution and the networks will allow us to partner with leading researchers in the field who can guide us on how to use our data gold to train ML models.