Circles Learning

Two weeks into the pandemic, I found myself as director of Chile's Center for Educational Innovation, facing an unprecedented challenge: schools were closing nationwide. Amidst the turmoil, a clear necessity emerged—to equip as many public school teachers as possible for the sudden shift to online education.

In response, we forged partnerships with tech giants Google and Microsoft to craft online courses. They were ready in two weeks! These courses were not just relevant and timely but visually appealing, packed with engaging, concise videos, and interactive Q&A sessions with seasoned instructors. Everything was accessible on-demand, tailored for busy teachers seeking flexibility.

The initial excitement was palpable; teachers enrolled in droves, eager to learn. Yet, despite our efforts and their enthusiasm, a stark reality soon became apparent. Over 90% of participants were dropping out—a disheartening echo of the global challenge faced by Massive Open Online Courses, as noted by experts like Khoushehgir & Sulaimany, and Wang et al. in their 2023 studies. High dropout rates has been long recognized as a persistent problem for MOOCs, their Achilles heel.

High dropout rates are merely a symptom. Our hypothesis suggests that the main underlying problem with remote, low-cost, mass learning is the lack of genuine peer interaction, which significantly diminishes engagement. We define engagement as the time, energy, and resources students devote to activities designed to enhance learning.

Many scholars agree on the importance of human connections. Professors Justin Reich (MIT) and José A. Ruipérez-Valiente (University of Murcia) write “most established research suggests that human connections through advisors, tutors, and peer groups provide the most important student supports…”. The problem is these human supports are expensive, “they will push against lower tuition costs. MOOC-based degree providers may find that highly effective online learning for diverse populations costs about the same to provision as highly effective residential learning.” In fact, high price, non massive remote courses (e.g. coding bootcamps), often do offer lots of opportunities for social engagement and support, even when online.

Reich and Ruipérez-Valente speak of access to advisors, tutors and peer groups. We focus on peer groups specifically. But how do peer groups link to engagement? To understand this question we can look into the framework of online engagement in higher education developed by Redmond et al. (2018). The framework identifies six elements that are important for engagement: Social engagement, cognitive engagement, behavioral engagement, collaborative engagement and emotional engagement. 

The presence of classmates can impact all of these dimensions. Social engagement and collaborative engagement are self evident but, for example, metacognition (one indicator of cognitive engagement according to the authors), is often activated when trying to explain something to someone else. Upholding online learning norms (an indicator of behavioral engagement) is easier to do when you know someone you care about is expecting you to show up, and managing expectations (an indicator of emotional engagement) is often linked to making commitments, explicit or implicit, with peers.

Circles app transforms remote learning into a constellation of learning circles, groups of up to five people designed for active learning, mutual support and collaboration. 

On a first encounter with the app, it feels somewhat like a dating app or a video game where students get to match with peers they choose based on shared goals, schedules, interests and proximity (circle peers often prefer peers who are close by to meet in person). 

After matching, the app guides peers in each circle to agree on joining regular, often weekly, study sessions for the duration of the course. When coming into their meeting, students can find their assignments for the session in the app, which includes asynchronous contents and prompts to guide group discussions and activities. During the session, students often get to work on individual or group assignments which they submit through the app itself.

The app has a number of additional features that we have developed based on users’ needs, including (1) group chat, (2) in-built group video calls, (3) a tool to help peers distribute group roles (e.g. taking meeting notes), (5) notifications about upcoming deadlines and (6) a "leaderboard" showing the ranking of circles (based on assignment completions), among others.

Circles is currently a mobile app available for Android and iOS devices and will soon (3-5 weeks) also be available as a web app so students can access it in desktop devices.

In turn course administrators like teachers and teacher assistants, join our "admin app", a web app where they can set up assignments for students, monitor progress, see when circles are meeting, track which users joined or not, engage with students via chat or video, help out with matching (e.g. assign a student to a circle if they didn’t find one to join), access student submissions and give students feedback. Notably, instructors have an AI assistant for giving students feedback, which makes it possible to reduce the time required for giving feedback and hence offer more feedback to students.

The theory of change driving Circles is grounded in the power of peer-driven learning. By creating supportive communities of people who collaborate, share insights, and challenge each other, Circles can significantly enhance peoples' motivation to finish courses and improve opportunities for active learning, leading to higher retention and better learning outcomes. 

At the moment we have courses with around 90% retention. Learning itself is harder to measure but we are building a promising tool to help with that. Soon, we will be recording the conversations people hold within Circles (with permission) and with that, get a sense about the depth of conversions, collaboration and other drivers that have been linked with learning outcomes. Couple this with A/B tests of session class plans, also under development, and we will be able to run randomized controlled trials to isolate the impact of particular interventions (e.g. a specific protocol to foster a greater sense of belonging among group members).

Student app demo: https://www.loom.com/share/018...

Admin app demo: https://www.loom.com/share/586...

We serve adult learners. Last year, with Chilean nonprofit Fundación País Digital, we ran two teacher training programs in rural communities of Chile, Atacama and Nacimiento, for 30 and 150 teachers respectively.

Rural training options include low-cost online programs (e.g. self learning platforms or large/impersonal webinars) and costly small-webinar programs In-person training is often unavailable for remote communities due to the higher costs and time demands on instructors. All in all, rural communities face a large tradeoff between quality programs and price. Circles works around that tradeoff: we can offer low cost programs that also have the hallmarks of quality programs: active learning, intimate conversations, rich exchange among teachers, working on group activities, and lots of feedback from instructors. In Nacimiento, we achieved a 90% completion rate, with most schools near 95%-100%, and one at 50%.

The experience working with FPD made it possible for us to identify key user needs and develop, test and iterate quickly on new features. Now, with a much more robust product, we just closed our first large client, INACAP, the largest higher education institution in Chile, with 4 thousand teachers. We ran a free pilot for them in January to train 70 teachers and we are about to kick off 6 courses on topics including AI for education, Gamification in learning, and Design Thinking. They expect between 1000 and 2000 teachers to join this first semester.

Two key reasons why INACAP is hiring us is that (1) we can offer better pedagogy at a price much lower than the small-cohort model, (2) we address INACAP's scheduling constraints that limit synchronous sessions to only winter and summer breaks, when teachers’ availability aligns. With Circles, teachers themselves work out the best time to meet their circle peers, allowing INACAP to expand training year round.

Similarly, we just started a partnership with ChileMass, a Boston based nonprofit that promotes Chile-Massachusetts knowledge exchange. With their sponsorship and the support of the Ministry of Education of Chile, we will now run a course for 1.000 teachers across the country. We are excited to see ~250 circles of teachers who feel listened-to, get to know each other over many sessions together, and work on group projects–all with the flexibility of self learning and the social dimension of the small cohort-based model. It's unthinkable to replicate a similar experience over Zoom or other existing platforms.

All of the examples above are in the education sector, training teachers. Our background is in the education sector so we have found it easiest to start there. But we would say that Circles addresses the needs of adult learners in general. We dream of now targeting other large institutions, government or private, that need to train large groups of people. We have identified a need in companies with large front-line workforces (people in call centers, warehouses, customer-facing workers) so we are now keen on exploring that option.

We have been in the shoes of institutions that need to train teachers at scale. Sebastian formerly led Chile’s Center of Educational Innovation, training thousands of teachers nationwide. The idea of Circles came about through ideating ways to improve massive open online courses.

Personally, both of us, Sebastian and Aanu, loved our time in college, taking courses, feeling challenged, engaged, feeling a sense of belonging. It’s unfortunate that adults struggle to replicate this educational experience post-graduation. We have taken many courses online in platforms like Coursera, edX, udacity and udemy, and the experience is nowhere close to that of college. The contents can be top quality, but everything else that makes a good learning experience is lacking. 

We understand our audience, having also been teachers ourselves. Sebastian, a math teacher, has designed and run training programs for ten years.. His innovative approaches in the past have earned him recognitions including the Ashoka fellowship, being named a LEGO Re-Imagine Education Champion, and ultimately being asked to develop and run the Center of Educational Innovation of Chile. 

We developed Circles through user-centric pilots over two years with 2,500+ participants. Incorporating the perspectives of the students we aim to serve is central to our development process. We monitor activity and interview users as often as possible. We have conducted over 100 user interviews and will continue to do so, ensuring Circles is finely tuned to the real-world needs and challenges of the teaching community. 

We have learned so much about the process. Initial assumptions, such as expecting proactive engagement among matched course peers, were proven incorrect. Key adjustments, such as setting expectations for weekly peer meetings and automating scheduling to overcome initial communication hurdles, significantly boosted our retention rates. We also thought that the contents for each lesson would be static. Then we learned that in Circles we have a chance of learning about the quality of each lesson every time a Circle does the lesson, and improve it for future Circles (since they are each joining the app at different times). In order to accommodate this insight, we ended up rebuilding our whole backend and integrations with the front end. We had to redo everything basically and start afresh.That was a blow after a year into development but it was right. This inclusive approach ensures our tool is not just a technological solution but a platform that genuinely resonates with and supports the populations it is designed to serve.

Finally, we are also the right people to work on this because we are both technical founders. We have written the whole app ourselves. Aanu handles the mobile app and backend. Sebastian is in charge of the front end of our web apps, UX design, and prompt engineering (given his knowledge of education).

We are in the Growth stage. We are available in more than one community and ready to scale. We have a consistent design approach (Of course, it keeps on improving and adding features, but that’s a constant for tech startups). We raised a seed round one year ago. We have served over 2,500 students. Now that we closed our largest client to date, INACAP, Chile's largest higher education institution, we expect to have over 3k active users this semester and will probably double that by the end of the year. 

Our current barrier is finding more partners to try and use our solution. We believe the exposure that we can get from Solve will help us reach new communities of learners, both in the teacher training space and other markets. As aforementioned, we believe that our tool is suitable for adult learning in general, not just teacher training and not limited to Chile. We would love to find partners in other countries that face the challenges of providing quality learning opportunities for adults at scale. If they are able to support our solution monetarily, all the better.

Also, we are interested in monitoring and evaluation support to build an impact measurement practice. In this regard, we believe that there may be academics interested in partnering with us to explore questions related to social and learning sciences that can be answered with Circles. We believe that Circles offers a prime opportunity to conduct randomized controlled trials at very low cost. Imagine that in Circles we are running the same lesson over and over again for each circle. In a course with 1000 people, there are already around 250 circles of learners, and thanks to LLMs we can now record many of those meetings and analyze them. You do not typically get a chance like that elsewhere. In traditional learning, be it online or in person, you are not likely to repeat the same class many times over a short time period. Moreover, we are now developing the capacity to provide slightly different session instructions to different circles. In other words, we will soon be able to A/B test things like the instructional approach of a given topic, or answer questions such as “Is it possible to forge an intimate friendship or sense of belonging quickly?”

1. We make online learning social. Current tools (ie. forums and peer-grading) used by mass learning platforms fail to generate meaningful conversations, much less meaningful social connections. 
2. Circles combines the flexibility of remote learning with the essence of in-person learning. When using Circles for large online courses you will always be able to find a handful of peers with whom you want to engage, that possess similar goals as you do, that have compatible schedules, similar background knowledge on the course subject, etc. You have synchronous meetings with asynchronous learning contents.
3. People learning in Circles are not necessarily online! People can choose to match with others nearby and meet in person. Nonetheless, we have all the data of whether they met or not, and what they did. This is why we speak of remote learning and not online learning.
4. We use AI to make the most of human supports, not replace them. Instructors leverage an AI integration to provide mass feedback. We are also building a tool to use whisper to learn from Circles' meetings and learn how to improve our lesson plans (Similar to TeachFX but using OpenAI's Whisper).

Problem Statement:

The primary issue is the high dropout rates and low engagement in remote learning environments, particularly in Massive Open Online Courses (MOOCs), which suffer from minimal social interaction and peer support.

Intervention (Activities):

• Formation of Learning Circles: Circles app matches adult learners into small, dynamic groups based on shared goals, schedules, and interests. This facilitates intimate and supportive learning environments.
• Regular, Structured Meetings: Encouraging consistent interaction through scheduled study sessions enhances commitment and accountability among participants.
• Integrated Learning Tools: The app provides group chat, video calls, role assignments, and other collaborative tools to facilitate active participation and resource sharing.
• AI-Assisted Feedback: Instructors utilize AI to provide timely and constructive feedback to learners, thereby enabling frequent feedback at scale.

Outputs:

• Increased Engagement: Regular and structured interactions within small groups lead to higher levels of cognitive, behavioral, and emotional engagement.
• Lower Dropout Rates: Enhanced engagement and support from peer groups result in significantly lower dropout rates compared to traditional MOOCs.
• Improved Learning Outcomes: Active learning and peer collaboration, facilitated by the app, lead to better comprehension and retention of course material. 

Outcomes:

• Enhanced Educational Access and Equity: By providing a low-cost, high-quality learning experience, the app makes education more accessible, especially for adult learners in remote or underserved communities.
• Professional Development and Career Progression: Learners acquire skills and knowledge that are directly applicable to their professional lives, leading to career advancement and economic improvement.
• Creation of Supportive Learning Communities: Long-term, the app fosters a culture of collaborative learning and continuous professional development.

Evidence and Support:

• Research on Engagement and Learning: Studies, like those by Redmond et al. (2018), highlight the importance of social, cognitive, and emotional engagement in online learning success.
• Internal Data: Out programs with Chilean teachers demonstrated a 90% course completion rate, significantly higher than typical online courses.
• Feedback from Users: Over 100 user interviews have provided insights into the effectiveness of learning circles, influencing continuous improvement and adaptation of the app.

Future Evaluation Plans:

To further validate this theory of change, we plan to implement A/B testing of instructional methods and use AI transcriptions and speech analysis to monitor the depth of conversation and collaboration within learning circles.

Impact Goal 1: Increase Completion Rates in Remote Learning

• Objective: Achieve a course completion rate of 85% across all remote learning programs.
• Indicator: Proportion of learners who complete their courses relative to the number of learners who enroll.

Impact Goal 2: Increase Engagement in Remote Learning

• Objective: Elevate the average depth of conversation in learning circles, aiming for at least 75% of discussions to meet predefined criteria for 'depth' and relevance.
• Indicator: Percentage of learning circle discussions that achieve a 'deep conversation' score based on analysis of session transcriptions.

Impact Goal 3: Increase the Use of Active Pedagogies in Remote Learning

• Objective: Ensure that 80% of all activities utilize active learning pedagogies as evidenced in session interactions.
• Indicator: Proportion of courses where session transcriptions reflect the use of active pedagogical strategies.

Impact Goal 4: Enhance Access to Quality Education for Adult Learners

• Objective: Expand the reach of the platform to include at least 50,000 adult learners annually across multiple geographic regions.
• Indicator: Number of adult learners engaged on the platform annually, disaggregated by geographic region.

Impact Goal 5: Promote Inclusive and Equitable Learning Opportunities

• Objective: Achieve a demographic parity index of 1.0 for enrollment across genders and rural/urban learners.
• Indicator: Parity index (female/male, rural/urban) for enrollment in all offered courses.

Mobile and Web Applications for students: The student-facing Circles app is developed using Flutter for mobile (iOS and Android) and React for the web application, ensuring a seamless and responsive user experience across all devices.

Admin Web Application: Utilizes React and integrates AI functionalities via OpenAI’s API and LangChain. This enables features like AI-powered feedback assistance for educators and speech analysis tools for evaluating learning circle interactions.

Microservices Architecture: Our backend, supporting both mobile and web platforms, operates on a serverless microservices architecture hosted on AWS. This includes AWS DynamoDB for databases, Lambda for event-driven code execution, Cognito for user authentication, and EventBridge for service orchestration, among others.

Supporting Technologies: Integration of third-party software for enhanced chat and video call functionalities.

Full time (2)
- Aanu Olakunle (CTO, full-stack dev)

- Sebastián Marambio (CEO, front-end dev and lead UX designer)

Contractors (3)

- Victoria Marambio (Instructional designer)

- Cristian Muñoz (UI/UX designer)

- Mohammed Ali Khan (Front end web developer)

2 years, since march 2022.

At Circles, we recognize the strength that diversity brings to our team, which is inherently multinational and multiethnic, with founders Aanu from Nigeria and Sebastián from Chile, and contractors from Chile, Colombia, and Pakistan.

Our commitment to diversity, equity, and inclusion is reflected in the following practices:

• Inclusive Hiring: We focus on skill and potential in our recruitment processes, actively avoiding biases that favor certain cultural or personal backgrounds.

• Flexible Work Policies: We accommodate diverse geographical, cultural, and personal needs, ensuring our team can work effectively regardless of location.

• Open Dialogue: We encourage a culture of constantly giving and receiving feedback about both technical work and our personal styles. We often have conversations where one person tells another about something they might have done that did not feel right, for example.

• Balanced Representation: We are committed to gender parity in our course offerings, ensuring diverse expert representation. For instance, in an upcoming AI course (www.chilemassonline.com), despite the initial difficulty in locating female experts, we achieved equal representation by proactively engaging qualified women to balance the roster.

Our key customers are institutions that need to train a large number of adult learners. At the moment, we are focusing on institutions that specifically need to train teachers, although we are already in conversation with a potential US client that would like to try Circles for training front-line workers. "

Our business model is B2B, charging per course seat (per learner per course). In the case of semester-long courses (10-12 weeks, with 1.5-3 hours of learning per week), the rate ranges between USD180 to USD40, depending on the number of learners. We charge additional fees for added services like creating the course contents or adapting existing courses of our clients, often to transform self-learning courses into courses apt for collaboration.

We provide the training service directly rather than the software. We onboard users, follow-up with them when they miss sessions, and give them feedback on assignments. We issue digital certificates and provide our client (the institution) with weekly updates on learners’ progress. 

As such, our business model is similar to those provided by corporate training companies: Coursera, Inc., Degreed, Inc., LinkedIn Corporation through its LinkedIn Learning services, Pluralsight, Inc., Udacity, Inc., and Udemy, Inc.

Again, the reason why our customers prefer us rather than such competitors is because of the unique social and active-learning experience we are able to offer for a price that is similar to that of self-learning courses and a fraction of what small-cohort courses usually cost (order of magnitude between US$ 1,000 and US$ 10k).

Last, we recognize that all of the competitor companies mentioned also offer services direct to consumer/learner. We have looked into that possibility and even built a customer-facing app to promote that (www.skillcircles.org) but put that project on hold until we have the manpower to tackle the challenges and investment that come with finding individual users, such as building a strong presence in social media.

Our business strategy for achieving financial sustainability primarily revolves around a B2B software sales model. Here are the key components of our strategy:

1. Target Market Expansion:

   • Educational Institutions: We initially targeted large educational institutions in Chile, leveraging our deep understanding of their ongoing training needs. This has proven successful with our significant contract with INACAP, which has the potential to expand significantly.
   • Corporate Clients: We are now broadening our reach to include companies with large workforces, particularly those requiring essential soft skills training for front-line workers. Our approach is uniquely suited to these needs because it emphasizes peer interaction, which is crucial for practicing and mastering communication and other soft skills.

2. Geographic Diversification:

   • We are actively seeking clients in both Chile and the United States, aiming to diversify our market presence and reduce dependence on a single regional economy.

Evidence of Success:

Our financial sustainability plan has already shown promising results:

• Investments and Grants: We have secured $52K from angel investors and $30K in grant funding, which supported our initial development and pilot phases.
• Revenue Generation: Our early collaborations have been financially beneficial, including $15K from a design partner and a significant contract with INACAP worth $32K for training 500 teachers, with the potential to increase to $100K annually as we expand our services to more of their staff.
• US Market Entry: We are negotiating a paid pilot with PathStream, valued at $10K. PathStream’s clients, including Amazon, are looking to improve their warehouse employees' soft skills, demonstrating the applicability of our software beyond traditional educational environments.

Financial Management:

• Current Burn Rate: Our monthly expenses are currently $6,000. This level of expenditure is sustainably covered by our current income and investment, positioning us well for future growth and expansion.