Agavi

A context-aware digital learning system for smartphones in low-bandwidth, low-power environments

Educational technology (edtech) is overly complex, too expensive, and uncreative.  During the COVID-19 pandemic, the transition to digital education left disabled, rural, and impoverished students behind because existing edtech systems are targeted towards privileged customers.

Agavi is a system designed for use by teachers and students everywhere else in the world, where bandwidth and electricity are often unreliable or unavailable.  In addition to low resource demands, the system is also designed to make science learning experiential.  Rather than being yet another system for showing videos and quizzes, Agavi allows teachers to utilize phone sensors and GPS location in their activities, making the smartphone an agent in its environment rather than a mere portal to passive content.  Additionally, the system is being designed with artificial intelligence capabilities, using activity performance data from around the world to help a teacher adjust content to be locally relevant and effective.

A key problem in many of the more remote parts of Indonesia is lack of connectivity.  With recurring waves of the COVID-19 pandemic, many students have had to withdraw from school and return home, which can include environments with low or no internet connectivity.  Even within the school ecosystem itself, connectivity is not always a given and can be expensive.  With much learning transitioning to high-bandwidth online modalities (ie, Zoom), this has left a lot of students excluded from continuing their studies.  The root problem here is a mismatch between tool and environment.  This problem stems from a skewed view of the state of global internet technology due to many edtech entrepreneurs existing solely in a high-bandwidth bubble.  They are unaware just how inapplicable their solutions are to most global teaching environments, which includes hundreds of thousands of students.  This mismatch is what our solution addresses, by developing edtech that is designed specifically for low-bandwidth, smartphone environments.  But we want to move beyond simply relaying content to students, which is the worst way to learn.  Our platform is designed to use the smartphone as an agent in its environment to make learning more place-based and experiential.

Our solution serves teachers and students in Indonesia, particularly those working in low income/electricity/bandwidth environments.  Many of these teachers and students have poor internet connections and older devices, ruling out the use of much edtech that has been used as a stopgap during the COVID pandemic.  While working in Indonesia as COVID was breaking out, I was able to personally experience these inequities and see how much they limited my teaching, as well as how poorly students were served by these edtechs.  This became more salient after I returned to the US and was able to quickly work around the bandwidth problems I experienced in Indonesia, but at the expense of making my activities inaccessible to students.  In response, I started a team to begin building an edtech platform that would work in the environments where I struggled with edtech the most.  As we are developing the Agavi platform, we are regularly engaging with our teacher partners in Indonesia to discuss their needs, observe how they interact with the system, and where they struggle.  This helps us drive the system to the least complexity possible in order to help improve uptake by teachers, as overly complicated edtech systems has been noted as a significant barrier to widespread adoption of edtech.

Our solution aligns with multiple aspects of the Challenge.  In particular, we align with access for low-connectivity regions, strengthening STEM skills, and supporting educators/leaders/stakeholders through adaptive learning.

Agavi is an adaptive learning system designed for the smartphone, which is a common and widespread internet access tool in the developing world.  Both the teacher and student interfaces are optimized for the smartphone.  The system is designed to have a low bandwidth footprint so that it can be easily utilized over poor internet connections.

Additionally, the system is designed to utilize a variety of inputs to help engage students.  Typical ones include multiple-choice questions, drop-down menus, and numerical inputs.  However, Agavi will go farther, including smartphone sensors and GPS locations as valid ways for students to input information into Agavi in response to teacher queries.  This will allow the development of new kinds of activities ... ones that incorporate a student's physical environment.  There's no better way to engage STEM learners than to have them explore their environment, report on it, and get feedback on their observations and hypotheses so that they can improve those skillsets.

Finally, the system will allow for adaptivity so that the tasks that teachers design for their students to complete can be linear or branching, responding to student answers and skills.  We are working to develop both computerized learning objectives to help teachers understand their students' abilities, and an AI-driven system that will look at global results and help teachers better localize shared activities and lessons.

Our team leader is Dr. Lev Horodyskyj.  He has been working in the education sector for 20 years, and with edtech for 10 years, including design, development, distribution, teaching, and education research.  He has experience living and teaching in the US mainland, US territories, and eastern Indonesia, and will be adding Ukraine and Brazil in the next year.  He is trained in astrobiology research and education at the PhD level, and as a result, excels in making unique linkages between people, concepts, and cultures.  Currently, he is living and working in challenging teaching and learning environments to gain first-hand experience in the challenges these communities face, in order to better inform the design of the nonprofit's programs and endeavors, including Agavi.

Agavi is innovative because it is being built to solve the problems of the most difficult learning environments first (this was driven by my teaching experience in eastern Indonesia prior to the COVID-19 outbreak).  Most edtech starts its life in high-bandwidth, high-tech environments, which results in design choices and architectures that are impossible to later modify for low-bandwidth environments.

Agavi is also innovative because it will treat the phone as an agent in its environment, rather than a portal to videos, text, and other passive content.  With the ability to access phone sensors (compass, barometric pressure, accelerometers, thermometers), wi-fi/bluetooth sensors, and GPS location, Agavi will allow teachers to build custom activities that have students using their phones like Star Trek's tricorders, something that is not currently done.

Finally, through the unique data architecture surrounding the tasks, Agavi will be able to identify and track "mutations" in content as it is shared and modified by other teachers.  This will allow research into how teachers adapt content that has never before been possible.  By identifying and harnessing these unique insights, Agavi will be able to identify common modifications and changes associated with specific learning environments, allowing other teachers to benefit from these insights as they prep content they find in Agavi's library for their unique classroom challenges.  Imagine having a system that automatically swaps continental examples for island examples if it knows that you live and work on an island.  Currently, that all needs to be done manually by every island teacher.

We are in the process of testing Agavi with partners in Indonesia, Brazil, Ukraine, and US Virgin Islands.  Currently we have given access to select teachers to observe how they interact with the system and what they build.

The core technology for Agavi is web-based, with app functionality planned for when necessary to work offline.  We are not interested in using anything cutting-edge, rather, we want to focus on "lowest common denominator" web technology so that we can have the broadest impact.  The less power and bandwidth that it uses, the better.

The novelty will lie in the low-cost tech ecosystem that surrounds Agavi.  For example, Agavi will be capable of running on SolarSPELLs (solar-powered Raspberry Pis distributed in impoverished areas to act as digital libraries).  Agavi's sensor ecosystem will also have an API to enable a "maker" community to develop around Agavi that we hope will build and extend Agavi's capabilities with sensors and tools that we haven't conceived of but that Agavi will, nevertheless, be prepared to handle.

Teachers have been stuck with poor edtech solutions for so long that they have an expectation that any new edtech will simply be a variation of the past 20 years.  They often have difficulty articulating what exactly they want because they have gotten so used to learning management systems, videos, and quizzes that that is the only way they can conceive of online education existing.  When asked to imagine differently, they leap to VR, AR, and other high-tech solutions that have a lot of novelty, but not a lot of practicality.  There is a gap in thinking of what may come in-between.

Agavi will take a step sideways, creating a system that prioritizes simplicity and novelty.  Simplicity will yield an easy-to-use interface and low costs of use (in terms of time, money, bandwidth, electricity, and equipment).  Novelty will yield new possibilities for activities that utilize coordinates and environmental sensors.  Together, these will empower teachers to quickly create imaginative interactive activities, and entice them to continue doing so as a result of the simplicity of the system.  This will have downstream effects in classrooms by making it easier to incorporate active learning approaches, which should consequently engage students.  Because teachers and students globally will be interacting within the same digital ecosystem, this allows the identification of common motifs and patterns that can be harnessed to help teachers connect with and benefit from the work of other teachers in the Agavi ecosystem.

My previous work with intelligent tutoring systems demonstrated the enthusiasm for such bottom-up empowering approaches, with strong demand from colleagues to have me train them in the use of an older ITS and share my content with them for reuse.  Unfortunately, the system was too complex and expensive to gain widespread adoption.  Our Agavi prototype has generated that same level of enthusiasm from teachers, who were already beginning to articulate ideas of what they would be able to do with it in their classrooms.  Teachers are ready for something different.  They have been for a while.

We are monitoring progress according to the available functionality of the system and enthusiasm/usage of our early testers.  We are also monitoring community interest in the project through Google analytics on Agavi's website, contributions to crowdfunding campaigns, conversations with strategic partners, and the number of new partners who approach us about the project.

Our impact goals for the next year are to begin selling low-cost subscriptions to a fully functional system and cultivating a dedicated user base.  We are currently working through volunteers who are building the first iteration of the system.  Through crowdfunding, bootstrapping, and partnerships, we anticipate being able to pull in enough funding to get us to launch.  Post-launch, we expect to grow via word-of-mouth.  Additionally, we will be developing strategic partnerships to gain users and add quality content to the network.  These will include working with the SolarSPELL project, which has a distribution agreement with the PeaceCorps for global deployment.  It will also include working with science education and outreach groups, such as the International Geoscience Education Organization, the International Astronomical Union's Office of Astronomy for Development, and the Blue Marble Space Institute of Science.  Each of these organizations works with teachers and learners at all levels, with a focus on teachers working in difficult environments.

We are addressing the lack of funds by working slowly and steadily, focusing on quality over speed.  Additionally, we are building partnerships with other organizations so that we can serve as vendors for their edtech needs.  Currently, we are developing a collaboration with the Etelman Observatory in the US Virgin Islands, as they have shown enthusiasm for the project and have a stronger record of winning grants that can be used to fund our initial efforts.

To address the challenges of working in the international marketplace, we have a board member at our nonprofit with business expertise who can assist us.  We are targeting specific countries where we have strong working relationships so that we work with just a few international markets at a time.

Regarding the scaling challenge, we believe that our approach of limited distribution to select partner countries will help us manage our growth so that it doesn't exceed our team's capabilities.

Science Voices was founded to improve equity in digital science education.  Prior to founding Science Voices, I worked at Arizona State University on a number of digital science education projects, but grew frustrated that we only paid lip service to equity and inclusion, often favoring expensive and computationally intensive simulators to teach science, knowing full-well that the materials would be inaccessible even within our home state, especially in Indigenous reservations.  We had been empowered with extremely helpful digital authoring tools, but the team chose to keep them internal and proprietary instead of training other teachers in their use.  With a focus on flash and a complete neglect of our digital infrastructure, the endeavor imploded shortly after I left.  This has driven me to try again, but this time with my own team that I can direct, actual lived experience in these challenging teaching and learning environments, and focusing on the basic digital infrastructure necessary to create lasting change in science education.

We have six people working on our team on a volunteer basis.  The lead is currently working in Ukraine (current location), Brazil, and the US Virgin Islands.  The back-end developer is located in Arizona.  The front-end developers are located in Romania.  The analytics experts are located in California and Pakistan.

The team lead has been working in the education sector for 20 years, and with edtech for 10 years, including design, development, distribution, teaching, and education research.  He has experience living and teaching in the US mainland, US territories, and eastern Indonesia, and will be adding Ukraine and Brazil in the next year.  He is trained in astrobiology research and education at the PhD level, and as a result, excels in making unique linkages between people, concepts, and cultures.  Currently, he is living and working in challenging teaching and learning environments to gain first-hand experience in the challenges these communities face, in order to better inform the design of the nonprofit's programs and endeavors, including Agavi.

Our back-end developer has 20 years of coding experience and hails from an Indigenous background and community.  He has firsthand experience with the challenges of education in Indigenous community settings, which informs the design of the Agavi's back-end functionality and user experience.

The front-end developers are enthusiastic Romanian students and recent graduates.  Growing up in post-Soviet Romania has given them deep insight into educational inequities and the patchiness of current edtech solutions.  Their recent experiences in education inform the design of Agavi's front-end functionality and user experience.

Our analytics experts hail from a disadvantaged backgrounds, having to work through difficult community and school settings to complete their degrees.  They have strong backgrounds in informatics, machine learning, and AI that will allow us build novel analytics into Agavi.

The team lead co-developed the Habitable Worlds online astrobiology lab science course at Arizona State University in the early 2010s.  He developed a number of simulators for star formation, inorganic carbon cycling, and cell metabolism (among a dozen others) to help students visualize and intuitively understand how the mechanisms behind these phenomena functioned.  He did this while coordinating teams in both the US and Australia.  He also worked on integrating these simulators into extensive lessons he built in an adaptive learning environment and evaluating how well they worked at meeting learning objectives.  Between designing, building, teaching, and researching this course, we was able to build the course from nothing into a highly successful course utilized multiple times a year at ASU to this day.  The course helped the team win numerous grants and awards from NSF, the Bill and Melinda Gates Foundation, and NASA, allowing the team to expand.

Blue Marble Space Institute of Science (WA, USA) - Science Voices has a memorandum of understanding with the organization for testing and distribution of Agavi globally as it grows

Etelman Observatory and University of the Virgin Islands (USVI, USA), Geneva Lake Astrophysics and STEAM Education (WI, USA), and the Ak-Chin Indian Community Library (AZ, USA) - we are working closely with these partners in developing Agavi to meet their institutional needs

University of Campinas and International Geoscience Education Organization (Brazil), Lviv Polytechnic and University of Lviv (Ukraine), and Khairun University (Indonesia) - we are working closely with select faculty at these international partners in developing Agavi to meet their instructional needs, particularly with hard-to-reach students in remote areas

More than funding, we need experienced guidance as we start on our business journey, especially as we expand our operations in Indonesia.  Most of us are trained teachers and programmers with little business experience.  We are learning as we are growing, but the challenges of growing and scaling call out for expert guidance, especially in navigating the international market.  In my previous job, I watched our intelligent tutoring system provider mismanage their business and eventually be forced to sell their technology to their competitor.  After the sale, teachers were cut off from the system.  I want to plot a more successful trajectory and implement the change we promise.

We are currently a small group of volunteers, whose previous experience has been teaching, scientific research, and programming.  We are confident that we can build an excellent solution.  We are less confident that we will be able to build a successful business around it.