Offline-First Learning: Ryan Ross on the AI Infrastructure Crisis

Offline-First Learning: Ryan Ross on the AI Infrastructure Crisis

Anika sat down with Ryan Ross to explore a critical, often-overlooked question in the education technology debate: Can your school's infrastructure actually run AI? The conversation revealed a sobering reality regarding the global digital divide and the massive data requirements of modern AI. Ryan’s background in edge computing and fintech at Visa offered a practical blueprint for offline-first learning, highlighting how Small Language Models (SLMs) and edge networks are the true key to bringing digital resources to the students the internet forgot.

In This Episode

• Transitioning from fintech at Visa to edtech and what education can learn from global payment networks
• The creation of the Olivia Education Edge Network and the power of offline-first learning
• Early deployment in East Texas and Hawaii, and the surprising "walled garden" benefit for classroom focus
• The impending AI infrastructure crisis: how token consumption and broadband limits will break school budgets
• Why Small Language Models (SLMs) solve the data sovereignty and policy compliance issues for districts
• Partnering with Utah Tech to deliver associate degrees to incarcerated youth inside secure juvenile facilities
• Scaling digital access across Africa and fixing internet-less computer labs
• Key takeaways from the UN's "AI for Good" summit in Geneva regarding global infrastructure disparities
• Practical, low-budget steps for educators to begin adopting AI this semester
• The guiding philosophy of putting "teachers first" in any technology rollout

Timestamps

• 00:00 Introduction: Can your school's infrastructure even run AI?
• 01:18 The Visa connection: Applying payment terminal edge networks to classrooms
• 04:18 Early use cases: Bridging the home broadband gap in East Texas and Hawaii
• 08:10 Scaling to enterprise and realizing the impending AI data crisis
• 10:27 Token consumption and why policy means nothing without infrastructure
• 14:49 Why Small Language Models (SLMs) are the safest path for school districts
• 16:57 The Utah Tech partnership: Bringing digital education into secure juvenile facilities
• 20:39 Deploying in Africa: The reality of scaling digital resources with minimal bandwidth
• 26:39 Why AI will never replace teachers and the realities of K-12 adoption
• 29:42 Insights from the UN "AI for Good" summit: The massive global infrastructure deficit
• 34:36 Practical steps for educators feeling behind on the AI curve
• 38:05 The guiding principle: Why successful edtech must put teachers first

Key Insights & Takeaways

Insight 1: Infrastructure Precedes Policy 

School districts and legislators can draft all the AI policies they want regarding data sovereignty and usage, but without the broadband to support it, the policies are useless. Global estimates discussed at the UN summit show that 80% of schools worldwide lack the infrastructure for viable AI, meaning the real frontier of educational technology is access, not algorithms.

Insight 2: Edge Computing is the Education Solution 

Drawing from Visa's global payment terminal model, offline-first networks allow schools to sync data locally. This middleware approach allows a school of 500 students to access digital resources, videos, and learning management systems on a localized network without crippling the school's limited live bandwidth.

Insight 3: The Hidden Threat of Token Consumption 

Adding generative AI to schools isn't just a broadband issue—it's a massive, unpredictable budget issue. A single school of 500 students could burn through up to 5 billion tokens in a year. Traditional OpenAI models are financially and logistically unscalable for the average, under-resourced public school district.

Insight 4: Small Language Models (SLMs) Offer Total Control 

Large language models pose massive data privacy and intellectual property risks for universities and K-12 schools alike. By utilizing Small Language Models managed at the district or state level, schools can strictly control the ingested content, perfectly align it with state standards, and safeguard student data in a closed environment.

Insight 5: EdTech Must Put Teachers First 

AI will not replace educators, because teaching is fundamentally rooted in human interaction. True AI adoption in the classroom will only happen when tools are explicitly designed to reduce educator workload, assist in curriculum alignment, and increase teacher productivity before being pushed to the students.

Resources & Links Mentioned

Olivia Technologies (Olivia Education Edge Network)

Utah Tech University (Juvenile justice education partnership)

UN "AI for Good" Summit in Geneva

About Ryan Ross

Ryan Ross is the founder and CEO of Olivia Technologies and the creator of the Olivia Education Edge Network. With a deep background in edge computing, fintech, and biometric authentication at Visa, he transitioned into edtech to solve the digital divide. His offline-first platforms bring critical digital learning and AI capabilities to students in under-connected environments—from rural American school districts and secure juvenile justice facilities to schools across Africa.

His core mission, Ryan notes: 

"Two projects I am particularly passionate about are Olivia's work with Utah Tech University supporting justice-impacted and incarcerated youth, and our efforts across Africa to improve access to digital learning resources for schools with limited connectivity. Both initiatives reinforce a core belief: technology should expand educational opportunity, not create new barriers. Whether supporting students in secure learning environments or schools that lack reliable internet access, our focus is on delivering equitable access to learning, empowering teachers, and ensuring that every student has the opportunity to succeed regardless of their circumstances or location."

Connect with Ryan

LinkedIn: https://www.linkedin.com/in/rossrr3/

Olivia Technologies Website: https://www.olivia.school/