Our Work
Examples of our recent projects
Maya - Chatbot Assistant to Help Combat Human Trafficking
Challenge: Every year, as much as 35,000 are trafficked in Nepal, the majority of them being women and children. Plan International aims to curb child trafficking in the country by educating teenagers about safe behaviours. Our challenge was to find a way to digitally engage Nepali teens with trafficking-related topics, augmenting Plan International multi-channel campaign PROTECT.
Our Solution: Together with Plan International USA and Nepal teams, we created Maya, a Facebook-based conversational assistant that interactively educates the teens on the topics of trafficking, migrating safely for work, and protecting yourself online. Maya utilizes quizzes and interactive stories to convey crucial messages of the PROTECT campaign and makes them more digestible for young audiences. Adventures utilized by the chatbot are written using real-world survivor stories and bring insight into human trafficking strategies employed nowadays.
Our Solution: Together with Plan International USA and Nepal teams, we created Maya, a Facebook-based conversational assistant that interactively educates the teens on the topics of trafficking, migrating safely for work, and protecting yourself online. Maya utilizes quizzes and interactive stories to convey crucial messages of the PROTECT campaign and makes them more digestible for young audiences. Adventures utilized by the chatbot are written using real-world survivor stories and bring insight into human trafficking strategies employed nowadays.
Predicting Underserved College Student Persistence With Data Science and Natural Language Processing
Challenge: Bottom Line helps first-generation students from low-income backgrounds get into college, graduate, and go far. In the organization’s signature programs, College Success, the students participating in the program receive consistent 1-on-1 support, advice and mentoring for up to 6 years. To improve the efficiency of the program, Bottom Line’s Data team wanted to identify which students are likely to drop out of the program
Our Solution: In cooperation with Bottom Line team, we have created a predictive model that identified student likely not to persist from year 1 to year 2 of the program, taking into account academic and program related data. In addition to tabular data, the model tapped into the counselor’s notes taken after 1:1 student meetings.
Our Solution: In cooperation with Bottom Line team, we have created a predictive model that identified student likely not to persist from year 1 to year 2 of the program, taking into account academic and program related data. In addition to tabular data, the model tapped into the counselor’s notes taken after 1:1 student meetings.
Enabling Sustainable Protein Production with Computer Vision - Powered Dashboard
Challenge: EntoLabs.ai aims to solve the protein crunch by leveraging the power of data to commercialize the insect protein industry. The aim of the project was to create an MVP showcasing the ability of machine learning to optimize insect incubation.
Our Solution: We developed a state of the art computer vision system to monitor insect colony health, count individual insects and help architect a system to optimize protein production rate and sustainability.
Our Solution: We developed a state of the art computer vision system to monitor insect colony health, count individual insects and help architect a system to optimize protein production rate and sustainability.
Empowering Microentrepreneurs in Myanmar With Personalized Content Recommendations
Challenge: make ONOW Myanmar’s Content Library accessible to serve thousands of microentrepreneurs using Sont OO financial Chatbot.
Our Solution: We designed and implemented a content recommendation engine that presented content to chatbot users based on their answers to the chatbot. The recommendation included both user tag-based recommendations, as well as natural language search capabilities.
Our Solution: We designed and implemented a content recommendation engine that presented content to chatbot users based on their answers to the chatbot. The recommendation included both user tag-based recommendations, as well as natural language search capabilities.
Improving Contraception & Abortion Education with Data-Driven Insights
Challenge: Understand the interaction of women in India, Kenya, and DRC with Nurse Nisa, a Whatsapp chatbot providing information on abortion and contraception.
Our Solution: We designed and implemented an analytics dashboard utilizing BigQuery, SQL and Google Data Studio to provide meaningful and actionable insight into the users’ engagement with the bot. In phase 2 of the project, we analyzed the free-text messages received by the chatbot and provided the client with insights regarding frequently asked questions, possible navigation issues, and areas for improvement, based on trends in natural language data.
Our Solution: We designed and implemented an analytics dashboard utilizing BigQuery, SQL and Google Data Studio to provide meaningful and actionable insight into the users’ engagement with the bot. In phase 2 of the project, we analyzed the free-text messages received by the chatbot and provided the client with insights regarding frequently asked questions, possible navigation issues, and areas for improvement, based on trends in natural language data.
Helping Female Entrepreneurs Beat Self-Doubt Through Science-Based Self-Empowering Chatbot
Challenge: Impostor Syndrome is a psychological pattern in which a person doubts their skills and accomplishments and has a persistent fear of being exposed as a “fraud”. Helping minorities and women understand and control impostor thinking is crucial in increasing diversity in the workforce. In this in-house project funded by VitalVoices, we aimed to scale the tool coaches use to help you reframe impostor thinking, and make them accessible as an interactive conversational assistant open to all.
Our Solution: We developed Syndee – an artificial assistant that can give you immediate 24/7 support to help you restore your confidence. Syndee helps the users to deal with thoughts that hold them back using scientifically-backed psychology exercises. These exercises allow the user to identify unhelpful patterns in their impostor thoughts, and reframe the self-sabotaging perceptions.
Our Solution: We developed Syndee – an artificial assistant that can give you immediate 24/7 support to help you restore your confidence. Syndee helps the users to deal with thoughts that hold them back using scientifically-backed psychology exercises. These exercises allow the user to identify unhelpful patterns in their impostor thoughts, and reframe the self-sabotaging perceptions.
Improving Student Focus With Behavioral Science-Based Mobile App
Challenge: Hour25.ai is a behavioral focus training program created to give college students the tools they need to stay focused, excel in their studies, and improve their wellbeing. The team needed to create a prototype that will be accessible through an Android, iOS and Web app, on a tight schedule and limited budget.
Our Solution: Within 6 weeks, we created an interactive, mobile-friendly Bubble app designed to help the users track their focus and perform media-rich interactive exercises. We also created an extensive data collection infrastructure to record every user action using Segment, BigQuery and Amplitude.
Our Solution: Within 6 weeks, we created an interactive, mobile-friendly Bubble app designed to help the users track their focus and perform media-rich interactive exercises. We also created an extensive data collection infrastructure to record every user action using Segment, BigQuery and Amplitude.
Helping Small Farmers in Myanmar with Natural Language Understanding
Challenge: enhance the capabilities of ONOW Myanmar’s GAP chatbot to understand natural language requests and direct the users to the appropriate information.
Our Solution: We built an English and Burmese Natural Language Understanding (NLU) module and designed a custom chatbot module in Botpress to integrate bilingual NLU into GAP chatbot.
Our Solution: We built an English and Burmese Natural Language Understanding (NLU) module and designed a custom chatbot module in Botpress to integrate bilingual NLU into GAP chatbot.
