SmartPal made it easy to:
Take existing art assets from books, textbooks, or other learning materials and make engaging and adaptive learning experiences that melded into existing stories or learning experiences.
Create highly varied interactivities to collect data about users and their skills, simulataneously bringing concepts to life, assessing the player's comprehension or learning from the interactivities and any accompanyting learning materials, and providing data to content owners and schools about the user's of their games and their content.
Quickly deploy interactivities and customized video games across any browser and into touchscreen apps using templated interfaces and game designs, including a fully customizable app portal that content owners could skin to match their brand.
SmartyPal mostly pursued content owners like publishers because it allowed games to be built into assets that were quickly being digitized at the time. It also allowed content owners who were not yet in the digital "game" to affordably deploy new digital assets quickly alongside their other content like physical books or products (e.g., toys and games).
The founders were initially focused on building interactive children's books for the iPad and they had learned that products aimed at improving STEM skills were drawing interest from schools, investors, and the US government. They are importance for preparing learners for technical and lucrative careers.
This was becoming a priority of the US government for allowing them to maintain a competitive advantage economically, due to evidence that the US was (and still is) slipping in international rankings associated with STEM skills.
The founders wanted to ensure their games were following learning science and would be effective.
I had a background understanding the importance of learning materials being in the zone-of-proximal development for optimized learning. I voiced concerns about whether certain activities in their e-book system could be verified to be at the right level to match the other material in the book and what would happen if the child using the book was too young or too old for the material.
I was especially worried that publishers using their system might not test the materials and would guess at the age appropriateness of content and the likely age of playing children. Based on my prior experiences with children's books, games, and commercial learning materials, I considered it very likely most interactive content would not closely match the child's knowledge and skills.
With one of the company founders, Kartik Hosanager, being a Wharton professor having a background studying recommender systems in business, we incubated the idea for SmartyPal's adaptive learning engine. It would use a learner's skill and the known difficulty of items to appropriately "recommend" that they interact with appropriate learning content.
We wrote the small-business innovation research (SBIR) grant together for a National Science Foundation (NSF) program. The grant offered the opportunity to fund R&D of the adaptive system with non-dilutive capital. At the time, the adaptive system was an innovative and risky approach that was difficult to execute with current technology. Traditional venture funding would have increased the need to quickly focus on creating return-on-investment rather than ensuring the technology's true efficacy. Working with the NSF as a partner would also open doors to additional relationships in the STEM and education industries and help establish the technology's legitimacy among educators and professional learning designers.
Phase I of the grant was funded in 2015, as shown here in Wharton Magazine with the early story of SmartyPal. I continued to advise the company and helped author and apply for Phase II of the grant, which was also funded in 2016.
Once I joined SmartyPal as an employee I led user and client research, product design, and led a product team including a visual designer, game developer, and engineer. We were now building what had morphed over the years from a simple digital book concept into a plan for a cross-platform adaptive learning system that adjusted content delivery based on real-time learner mastery. I would use my knowledge of the education industry and prior school partnerships, research skills, and learning science and technology-mediated background to continue to move the platform from proof-of-concept to a viable product.
Working with the founders, a salesperson, and potential and pilot clients, I collected information about what customers needed, informing our roadmap. I then worked to build those things into the system. For the product to be viable for B2B users like publishers and schools, we had to build out for many administrative functions. These included a web portal and analytics dashboards for educators, parents, and publishers to monitor skill growth, track learning outcomes, and identify good and poor performing learning content.
I also continued to improve and test the efficacy of the product and algorithms in ways that would help make it more viable as a product. For example, we had to tune the algorithms to begin adapting to player skills or item difficulties faster when these were not well established in newly created games or among new users.
Intimately understanding both the architecture of the system and the learning science behind it, along with having established education-related credentials, I regularly represented SmartyPal's brand as a serious learning company in important business meetings, at conferences, and in other promotional initiatives. I also also created business development materials.
A Unity-based template system for creating cross-platform interactive content and mini-games at a fraction of the cost of building standalone customized interactive content.
A system of customizable SmartGames that enabled customers’ to even more easily incorporate content into fully functional mobile-style learning games.
Interfaces and a web portal for showing learners, educators, and content managers how users were performing on their content.
A cloud-based Adaptive Learning Engine that integrated with the templating and SmartGames system, ensure the relevance and efficacy of the presented content.
The template system was mostly offered to academic publishers and corporate partners with large libraries of static, boring content that could benefit from being transformed into interactive experiences. For example, a mini-game for a book on bats in which the user had to find food using the principle of echolocation.
Noting some potential customer hesitance to take on significant design requirements or not needing to do deep customization to take advantage of gamified practice and assessment (e.g., university teachers or certain corporate learning professionals), SmartGames took the template system a step farther to offer fully functional and tested games that creators could simply insert their questions and content into. The SmartGames did not offer the same customizability, but could be created and deployed in minutes rather than hours while still being able to access the benefits of the adaptive learning system, ananlytics, and much of the fun of game-based learning.
I also created the curriculum and games for KidsMakeItRight:
KidsMakeItRight was an after-school program powered by SmartyPal. Our founder and I did it as a combined user experience research opportunity, a way to build partnerships with local schools, and a passion project to create impacts of our technology in the local area, where cleaning up the environment was a city-wide priority for Philadelphia.
SmartyPal was used in the curriculum to "dogfood" our own technology for use in K-12 institutions. Trying to quickly create games and then use them in a real classroom while managing 10+ real children gives you a very different perspective on whether your technology "works" and what it needs to enable learning at scale.
Though the company started as an e-book creator for young children, the adaptive engine and template system was built for any content for any age and games could be created to adapt to find the appropriate difficulty levels for very large age ranges. The games and any other learning materials simply had to be built to support large age ranges. The company increasingly pursued game-based learning applications with older audiences where repeated practice and more careful assessment was often necessary.
The Adaptive Learning Engine powered a Lietner Box-inspired recommender system monitoring how users were performing on content and what content they interacted with recently. It tracked user performance, the difficulty of items (which it automatically learned), and the time since a user interacted with that content to space out learning and monitor ongoing mastery.
The learning system was all aimed at making the learning process more efficient no matter your age or learning content. The system ensured:
Content was in the "goldilocks" zone; not too hard, not too easy.
Directed toward relevant content that an individual needed to learn or be refreshed on.
Spaced so that retention was maximized and information was not just memorized quickly and forgotten.
Built to make the learning context more active, engaging, and to reduce frustration relative to static learning materials.
One of the hallmarks of the system, as shown via the solid black line in the image below, was that it closely spaced repetition of new, lesser known content with other newer content until the newer content was "mastered". Once, mastered, the system would display that content less frequently but still periodically repeat content to check whether learners had forgotten.
The dotted line shows the typical approach where a learner may be given a learning module to complete in a day, ramped up to high levels of knowledge, and then left to forget that information until a 6-month refresher course. Given how human knowledge decays quickly, especially after massed learning, that approach requires more total learning time and yields much longer periods of time where the learner would have unacceptably low performance if assessed.
Spaced repetition is an efficient way to learn anyway, this system allowed content managers to simply add new materials to an existing library. The system ensured that anyone using the system would be presented the content often until it was mastered. The system plus online portal could then also give those content managers constant reassurance that learners were maintaining acceptable knowledge of all of the materials.