Afinidata

250 million children under five fail to reach their potential in cognitive development.

Research has shown that home-visit intervention to parents for early-childhood education (ECE) has long-term impacts on child development and their future prosperity. However, home visits to millions of families is simply not scalable. Afinidata combines the proven model of home-visits with the power of artificial intelligence to provide a solution to ECE that is cost-effective, scalable, and does not require behavioral changes from parents.

Within social messaging apps parents already use to communicate (Facebook Messenger), Afinidata offers a free virtual assistant that reaches out to parents, learns about their children's development, and recommends personalized ECE activities.

We are currently serving over 25,000 parents, primarily in developing countries. In 2019, UNICEF Innovation Fund selected Afinidata as one of the most innovative companies around the world with the potential to impact over 1 million lives.

Millions of children around the world under six years do not have access to quality ECE to be ready to start school. For example, in a country like Guatemala only 1.3% of children have access to ECE resources. As a result, these children will be delayed by the time they enter primary school and are 25% less likely to graduate from high school. Unfortunately, most parents in developing countries lack the time and resources to access quality ECE programs.  Governments and NGOs have been trying to improve access and reduce costs of their ECE home-intervention programs; however, despite positive outcomes of these interventions, scalability is limited. Additionally, home-visitors often do not receive sufficient training nor do they receive adequate support while conducting home visits, especially in remote areas. Private solutions are expensive and not practical for low-income communities. Technology-based solutions can create accessibility issues by requiring full access to internet.  Text-based programs have positive impacts; however, they are limited to 160 characters, lack multimedia interaction, and in developing countries could require payment per message. Furthermore, most current educational activities are administered to all children of a given age; this lack of customization can reduce parents’ overall engagement. 

Afinidata was developed in low-income communities in Guatemala, a similar socio-economic and cultural setting within Latin America. Through community workshops we learned about parents’ specific needs for ECE as well as the availability and use of technology.  As our community workshops started to grow, we identified that a low-cost solution was needed that can reach millions of parents with relevant and personalized ECE content that will keep them engaged. Once we introduced Afinidata via social media messaging channels, we had hundreds of parents signing up within a few hours. Since then we have used all of our interactions with parents to continually enhance product development.  Currently we serve thousands of Hispanic families in developing countries, mostly located in Latin America (e.g., Guatemala, El Salvador, Honduras, and Chile).  Most of our users are first-time mothers who have access to a cell phone and social media. Parents using Afinidata find value in having access to a 24/7 ECE expert within the channels they already use to communicate, which increases the time they share with their child while helping them prepare for school.  Although Afinidata has been designed for parents, we currently have organizations and professionals using Afinidata to complement their home-visit programs.

Imagine if each parent had their own free personal 'assistant' to help them build the resources they need to provide adequate early-childhood education to their children. This is what Afinidata does! We solve ECE accessibility and content personalization issues. 75% of low- to middle-income families have access to Facebook Messenger. Afinidata offers a free, personalized, virtual assistant in Facebook Messenger that guides parents with ECE tools for children under the age of six.

Here’s how it works:

1. A parent subscribes to Afinidata by starting a conversation with our virtual assistant. Parents typically find us through teachers, health centers, friends, relatives, or Facebook ads.

2. Every week our virtual assistant reaches out to the parents and chats with them. Afinidata’s virtual assistant is able to learn and track a new set of variables about parents and children with every interaction. Rather than hoping parents will later engage in these activities, the virtual assistant observes whether they access the activities and solicits feedback.

3. Similar to ECE home visitors, the virtual assistant learns and incorporates parent’s feedback about their children’s well-being and using artificial intelligence makes suggestions for health- and development promoting ECE activities. In addition, parents can reach out to Afinidata’s virtual assistant at any time, asking questions, sharing special achievements of the children, or seeking advice.

All of our educational activities had been designed in Spanish using the Montessori framework with material that parents can find at home. Each activity has been designed to help children reach certain milestones in the following development areas: motor, cognitive, language, and social-emotional.

Artificial intelligence incorporated in social messaging channels millions of people already use is simply a game changer for providing an innovative and personalized solution to deliver quality ECE resources to millions of parents within seconds. Afinidata leverages the reach of Facebook Messenger, a channel that has 1.2 billion active users, which does not require new downloads and is inclusive with lower-income families. By using artificial intelligence and more than a million data points collected from Afinidata’s users, we have created a unique, highly-accurate algorithm that transforms how ECE content is recommended.  

Afinidata is a major innovation compared to traditional in-person home visits, and has the potential to fundamentally change government and NGO ECE programs by taking then to scale.  Compared to existing private and tech-based solutions, Afinidata is free to caregivers and allows integration of personalized multimedia content. Considering that we track child development, Afinidata’s virtual assistant can identify when additional expert consultation is needed for a given child. We are then able to increase accessibility of the professional services to families via chatrooms and video services or partnering with government and NGOs. Finally, Afinidata can also be expanded to globally with main effort required being translation and adaptations of materials.

Afinidata uses artificial intelligence, machine learning, and Natural Language Understanding (NLU) incorporated within messaging channels such as Facebook Messenger or WhatsApp. Our technology includes three main components: content library, virtual assistant, and content recommendation algorithm. The content library is a centralized database of educational content where different ECE experts can contribute. Each activity in the content library has specific metadata that helps classify the ECE content to be recommended to a specific caregiver. The virtual assistant is in charge of interacting with parents via messaging channels to identify the intent of the caregiver and provide adequate responses. The virtual assistant is trained to handle contextual dialogues with a deeper learning. During each session with the caregiver, the virtual assistant collects information about the caregiver, home environment, and child development:  including demographic information (e.g., date of birth, country), sociodemographic information (e.g., scholarity, occupation), home practices (e.g., how often they read to their child, how often they play together), and child development and progress. Lastly, the content recommendation algorithm uses the data the virtual assistant has collected combined with pre-trained data from more than one million interactions with our users to recommend ECE activities to parents. The recommendations are performed using artificial intelligence based on all the information we collect from each parent, prior use and performance of each activity, and their child’s specific development. 

The most successful model for improving child health and well-being in low-income settings is home-visit programs to support parents; such programs have been successfully implemented in dozens of countries and have proven to be effective. ECE home-visit programs, like Reach Up Jamaica, have led to better prepared students to enter primary school and a 25% increase in income based on participants surveyed 20 years after the home-visit program. Afinidata works because we use the proven successful model of home-visits coupled with highly-adopted technology that parents readily use.     

The rapid raise in cell phone coverage globally offers new opportunities to reach parents, dynamically collect user information, and use this information to optimally calibrate ECE services. Afinidata leverages the broad adoption of social media messaging channels to improve health and development of children under the age of six. In Latin America, telecommunication companies have created phone plans for lower-income families that include unlimited use of Facebook Messenger. Afinidata integrates and runs directly within this high growth and penetration app, to fill a need for effective, easy-to-access educational tools that are inclusive with lower-income families.

Afinidata’s virtual assistant is highly accepted by caregivers; over half of our users have recommended us to a friend and continue using us on a weekly basis. Furthermore, an experiment with 7,349 caregivers showed that our content personalization algorithm was statistically relevant (p-value < 0.001), whereby Afinidata-recommended content was higher among caregivers compared to age-based content suggestions, which are the norm for most available solutions. 

More than 25,000 families have subscribed to Afinidata. Although we have families residing in 10 countries, the majority reside in Latin America, including Guatemala, El Salvador, Honduras, and Chile. By the end of 2019, we expect to reach 100,000 families. In 2020 we will start operations in partnership with the Swiss Tropical and Public Health in Peru and also expand our presence in the U.S., focused on Hispanic families. By the end of 2020 we expect to reach 500,000 families. We believe that in five years we can reach a global scale and have more than 10 million families subscribed to Afinidata. In terms of adopting Afinidata to other continents, the main effort is translation and adaptation of our educational materials.  We plan to create partnerships with public institutions, content developers in other languages, and global educational programs to help Afinidata grow at the global scale.

Our goals over the next year include the following:

• Provide free access to quality educational tools to 100,000 families;
• Deliver 1 million free, personalized educational activities and tips through our virtual assistant;
• Release Afinidata’s virtual assistant version for professionals that can support and complement practioners conducting home visits;
• Grow our premium and government partnership business models to ensure sustainability of Afinidata; and
• Improve and optimize our content recommendation algorythm to create solutions that continue to be inclusive for lower-income families and provide high-levels of content personalization, that prove to increase child development by 2 months over a year.

Our goals over the next five years include the following:

• Conduct an impact analysis that demonstrates Afinidata’s solution is comparable to in-person home visits from ECE practitioners;
• Establish a global presence and reach 10 million families;
• Be adopted by at least four governments as the platform for assessing child health, well-being, and development;
• Integrate within health care systems that allows referral in emergency cases and alerts for vacinnation reminders; and
• Reach break-even with our business model. 

The barriers we anticipate over the next year include the following:

• Access to capital;
• Presence in media and overall exposure of Afindiata;
• Parents frequently approach us seeking specific advice in areas such as health, vaccinations, and nutrition. It has been difficult to gather trust-worthy content in these areas and partners that can provide more in-depth advice.
• Independent study that shows Afinidata’s impact. 

The barriers we anticipate over the next five years include the following:

• Dependence on a single messaging platform (i.e., Facebook Messenger);
• Limited physical presence and partners outside of Guatemala and the United States;
• Content translations to different languages and dialogues; and
• No globally-accepted standard for governments to measure child development. 

Actions to overcome barriers over the next year:

• We are applying to different global funds and preparing for a round of investments;
• Advancing different digital strategies and local partners to gain publicity;
• Partner with NGOs to create further interventions pertaining to child health and development when needed;
• Enhance algorithm to recognize messages from parents and redirect to experts with collaborative partners;
• Applied to a grant through the Botnar Foundation to assess Afinidata in Peru in parallel with home-visit programs.

The barriers we anticipate over the next five years include the following:

• To overcome concerns related to dependence on a single messaging platform, we will consider developing our own software or app to address the needed of our users Dependence on a single messaging platform (i.e., Facebook Messenger)
• Potentially develop in-country offices in areas where we considerable growth in the number of users.  These offices would have staff devoted to networking and support.
• Developing a platform that is interchangeable with development tracking tools. Also, we are establishing partnerships with organizations that conduct research in the area of child development to collaborate on tools.
• Plan to use a combination of machine learning algorithm translations coupled with partnerships in other languages. 

Not Applicable.

Full time staff = 3

Contractors = 4

Our team is focused in the following areas: content, marketing, data science, and product development.

Creating scalable technology-based solutions for early-childhood education requires multiple and diverse skill sets, such as understanding how children learn from birth to age six, coding and algorithm development, and business model evaluation. Our team has over ten years of experience in education, data science, and artificial intelligence. Most importantly, we understand our users: how they interact with different technologies, what content they find relevant, and how to reach them. We had devoted numerous resources to find the most cost effective way to enroll new users into our program and keep them engaged, achieving 70% retention rates after 90 days, which is one of the highest retention rates for ECE programs.

Afinidata works with some of the best technical talent in Central America. Andreana Castellanos, Founder of Afinidata, is an Industrial Engineer and Fulbright Scholar with an MBA.  Andreana has collaborated with a group of researchers from Harvard on their Caregiver Reported Early Development Index (CREDI), via implementation of CREDI to assess child development in three culturally diverse settings. The Afinidata teams also includes a lead developer who has worked for top companies like Google, and a lead data scientist who has a PhD in Physics and Math from UNAM in Mexico.  In addition, the Afinidata team includes educators that author and revise the educational content. Thus, Afinidata has the necessary expertise related to educational content, artificial intelligence algorithms, data assessment, and business development to continually broaden our impact on early childhood education.

Current partnerships 

1. UNICEF Guatemala - We have been working in partnership with the local UNICEF office in Guatemala to promote Afinidata and currently we are evaluating other routes of collaboration with the regional Latin America office. 

2.Guatemalan school districts, government programs, and NGOs - Through these programs we have reached new families who have enrolled in Afinidata.

3. Swiss Tropical and Public Health Institute - We are collaborating on independent research to conduct an impact evaluation of Afinidata compared to in-home visits. We are also expecting collaboration from the Peruvian Government with this assessment.

We are currently experimenting with two revenue models: (i) premium subscription and (ii) collaborative partnerships. 

The premium subscription model competes in a market industry of 15 billion USD, which is the market of parents with children between 0 and 6 years old in the U.S.  At the moment our target market is Hispanic parents in the U.S. and Latin America. We expect that from our initial pilots of the Premium model that three percent of Afinidata’s users will purchase the Premium version that includes unlimited access to educational activities and tracking of progress for their children. The monthly cost for Premium is $3 USD, which translates to a lifetime value of a Premium user of $72 USD (considering a Premium user will retain their subscription for two years).

During 2019 we have gained attention from international aid groups and government organizations, with whom we are exploring the implementation Afinidata’s virtual assistant to assist at-home visits from caregivers, educators, and other service workers. In the upcoming years we anticipate that Afinidata will be implemented as part of the healthcare system in at least two countries in Latin America. These potential collaborative partnerships will create a new source of income while dramatically transforming and reducing the costs of early childhood development for these institutions.

We expect a combination of grants, investments, and revenues to cover our expenses. We currently are applying to strategic grants, including this one. As we continue to grow, the potential use of Afinidata around the world is large. Most households have cell phones, and the use of messaging channels is growing at an unprecedented pace; for example, Facebook alone has almost 2 billion users globally now. In terms of adopting Afinidata to other countries, the main effort required is really translation and adaptation of materials, which could be achieved for minimal costs and generate considerable earnings.  We have been evaluating and testing multiple avenues to monetize our user database, including the Afinidata Premium model, ads and market research, and providing services to international aid groups and governments that complement their home visit programs. We not only expect these monetization models to cover expenses, but also to generate profits.

We share with Solve the vision that a complex problem cannot be resolved from one organization alone. We are interested in connecting with the rich ecosystem of innovators at MIT Solve, especially with those related to the solutions at the front-line of health. These relationships could help address one of the biggest opportunities / barriers in our business, which has been an increase in specific requests from parents regarding assistance with health-related questions. Parents are seeking specific advice in areas such as vaccinations, medical treatments, and nutrition. We have identified some MIT solver teams using technology to help community workers and health records that could enrich our model and provide a more holistic view of child development.

We will benefit from Solve partnerships and media exposure to continue scaling Afinidata. In addition, we are interested in partnerships and mentorship to enhance our machine learning capabilities and refine our business models. We see the Solve program as a unique opportunity to engage with a progressive community of innovative thinkers to help us advance Afinidata.

Addressed above.

• Botnar Foundation – We are applying for a social impact analysis grant to conduct a rigorous mixed-methods evaluation to compare in-person home visits with Afinidata’s virtual assistant intervention.
• The Bezos Foundation – They support a program called Vroom, an app through which tips and activities are provided to enhance ECE education. We will benefit to have insights on their content and experience.
• Gates Foundation – They are conducting ground-breaking work regarding access to vaccinations to promote global health. Vaccinations, reminders, and alerts for parents are part of Afinidata’s product roadmap.
• Johnson & Johnson – We believe companies like Johnson & Johnson could benefit from our technology, in particular from our insights, personalized content recommendation system, and relationships with parents. 

Afinidata is using the power of artificial intelligence (AI) to transform accessibility to quality early-childhood education (ECE) resources around the world. The AI Innovations Prize would be used to continuously enrich our virtual assistant NLU ability to handle parental requests, advance our machine learning capabilities, and optimize our content recommendation algorithm. Enhancements to our AI technology will maximize the potential impact of Afinidata. Our current algorithm will be improved through the latest machine learning technology towards the goal optimizing both the systematic collection of data and content recommendation algorithm for low-income families. We will also prepare the system for further cultural adaptation should the system be scaled-up to other countries or languages.

Afinidata’s mission is centered on providing a free, personalized, virtual assistant in Facebook Messenger that guides parents with ECE tools for children under the age of six.  We currently have more than 25,000 parents registered with Afinidata, with over ninety percent of them being first-time mothers. These women find value in having access to a 24/7 ECE expert within the channels they already use to communicate, which increases the time they share with their child while helping them prepare for school.  Although our mission has been enhancing child health and development through free ECE activities, we are helping women by promoting better connections between them and their children and helping them feel more confident and less stressed about their jobs as mothers.

We also connect with pregnant women, who are using Afinidata to be prepared to assist their children from the day the enter the world. If awarded the Women Prize, we will focus on enriching content we currently have available for pregnant women and self-care content for mothers. For example, we envision developing the abilities to provide reminders for health center visits and nutritional consultation, along with enriching our database for activities mothers can do with their child while still in the womb. Afinidata’s goal is to promote healthy babies, which starts with healthy, informed, and engaged mothers.

Afinidata provides free early-childhood education (ECE) content to parents such they can engage and educate their children in the comfort of their own home.  We view the application of AI in this ECE solution as disrupting the social status quo of relying on home-visit programs that can be an economic burden to scale.  We will use the Innospark Ventures Prize to scale Afinidata to 100,000 additional parents. We will enrich our virtual assistant, Natural Language Understanding, and content recommendation algorithm to integrate within governmental health systems that currently rely on home-visit programs for administering ECE. We see considerable potential for integration with existing government and NGO home-visit programs and believe that Afinidata can be a disruptive technology that enhances society via more accessible ECE while minimizing the economic burden.

The best way we have found to protect this data is to use the following steps: (1) store all data on a private server; (2) restrict access to the user database to select individuals with privilege; and (3) encrypt the data. For the select individuals that have access to the database, we make sure they are aware of the risks and responsibilities assumed with handling private identifiable indicators.