The Formation Crisis Part 2: Core Infrastructure for the Long-Term Solution

Open-source adaptive AI tutors, soulful human guides, and church-based pods where students lead the renaissance of their parishes—this is how we democratize formation without being locked into a billionaire's closed system.

"Write the vision; make it plain on tablets, so he may run who reads it." (Habakkuk 2:2)

I recently consulted for an AI-tutor-powered school system that's gotten a lot of attention. One thing I will forever be grateful for is the fact that this company has put adaptive AI education tutors into the zeitgeist. It costs a lot of money to put something into the zeitgeist and accelerate the time horizon. They accelerated the conversation around adaptive AI tutors by five years—maybe more. In the same way that Tesla accelerated the conversation around electric vehicles.

But here's my concern: I don't want to live in a world where a billionaire has a monopoly over adaptive AI learning tutors.

The Inevitability of AI-Assisted Formation

There's no question in my mind that adaptive AI is how all kinds of kids are supposed to be learning skills for life—reading, writing, arithmetic, and whatever else. The technology is here. It's only getting better. The question isn't whether AI will transform education. The question is: Who controls it?

In The Formation Crisis, I argued that the myth of separated formation—divorcing spiritual development from intellectual growth—is destroying our children. Secular schools teach atheistic ontology while parents outsource formation to systems that undermine their values. The solution I proposed there was modular, open-source formation tools and Christofuturist schools. This piece is about the core infrastructure that makes that vision possible.

What does it take to create a center of gravity around an open-source alternative? One that may take time to catch up, but once it does, compounds exponentially?

The Linuxification of Adaptive AI Learning

I don't know exactly how it would happen, but over time, I would like to see the Linuxification of adaptive AI learning infrastructure. A system where if someone wants to contribute a piece to the ecosystem, they can—and they can feel like it's not just energy that dissipates into the wind because it's part of a broader ecosystem with credibility and trust.

Think about what this means practically:

Modular Building Blocks: The adaptive AI formation infrastructure becomes modular. Different teams can build and refine different components—math modules, reading comprehension modules, biblical knowledge modules, civics modules—each improving through collective iteration.

School OS: Imagine a parent or microschool teacher prompting a "School Setup OS" with basic information: Where do I want to launch this school? What state standards apply? What are my family's values? The system automatically recommends which subjects to teach and which highly-rated adaptive learning modules to use for drilling kids on different concepts.

Faith-Based Defaults: For faith-based schools, the defaults for biblical knowledge formation are already configured, but you can always change which modules you use. Your values shape the journey.

Open Contribution: Anyone can contribute to this collective intelligence infrastructure. It's humanity collectively building systems that help every individual child access high-quality formation.

This isn't pie-in-the-sky. Projects like Hugging Face's Transformers library already enable modular, community-driven AI development. Imagine a "School OS" as a frontend to something like that—pulling from a decentralized hub of formation modules that anyone can fork, improve, and contribute back to.

We've already seen hints of this future. CK-12, MIT OpenCourseWare, Khan Academy—these projects proved that high-quality educational content could be created and distributed freely at scale. They democratized access to knowledge. What's needed now is the next evolution: not just open content, but open systems that personalize the learning journey while remaining transparent, customizable, and community-governed.

Why Open Source Matters for Formation

The leading AI-powered learning systems cost around $10,000 per student per year. That's very expensive. And from what I've been told, there's not an obvious open-source equivalent.

There are people trying to create software for microschools and homeschools. But capital-intensive, VC-backed companies have emerged as the default if you don't have strong opinions about what your kid should be learning.

This is the problem with closed-source formation infrastructure.

Parents who want to homeschool their kids should be able to put their own opinions into a homeschool journey creator and plug into meta-best-practices for how to educate their kid based on what's best for them—whatever that may mean for their family, their faith, their values.

It feels much more reasonable for a parent to communicate their values and have a highly trusted, open-source adaptive AI infrastructure—representing collective evolution—create a pathway for their kids, rather than being forced into closed-source infrastructure controlled by a single company.

See: The Truth Economy for understanding how Christ-aligned alternatives to centralized systems serve kingdom purposes.

The Winning Combination for Democratizing Quality Formation: Soulful Guides, AI Tutors

I've had interesting conversations with people who want to start microschools. Most of them are grossly underestimating how hard it is—even with the current state of adaptive AI tutors. Without them, it's virtually impossible.

One friend put it bluntly: "Why would I start a microschool? I'd have to be the principal, the teacher, the janitor, teaching all subjects at once. It'd be basically impossible without an incredibly good tech system." And to be fair to her point, even with an adaptive AI tutoring system it's still incredibly difficult—a single-point-of-failure where one leader could easily burn out and ruin it for everyone.

I'm not even saying microschools are completely the future. But AI tutor systems absolutely are going to be the default—even in larger schools. If we want more values-aligned formation systems for Christians and non-Christians alike, we're almost inevitably going to have to build on these adaptive AI systems. We better make them as good as possible.

Here's what's much more feasible for billions of kids in the future around the world: having a loving, soulful guide rather than finding an excellent math teacher, excellent civics teacher, and excellent writing teacher all in one place.

It's much more likely that children can find in their local area a great soulful, perhaps motherly figure that wants to see the best for them—while delegating the knowledge drilling to AIs that will keep getting smarter and smarter.

The synergy is symbiotic: humans focus on values, purpose, inspiration, motivation, love, support, and accountability. Smart machines focus on the core fundamentals—quickly assessing where a student is relative to where they need to be, drilling them on the basics, adapting in real-time. People can debate what those fundamentals should include, but the combination of soulful human guides and intelligent adaptive systems is both winning and inevitable.

So let's let these smart machines do what they're going to be increasingly good at. But let's also give them really good frameworks to plug into.

Tech won't fix everything, though. We also need social infrastructure—co-ops, networks, and communities for sharing best practices, preventing burnout, and ensuring no [micro]school leader is an island.

Strategic Community Building: Christofuturists as the Vanguard

Strategically, it makes sense to build communities at the intersection of highly opinionated communities and tech. Conservative Christians who are also deeply embedded in technology represent one of the highest-priority intersectional demographics to organize around. I call them Christofuturists—people who hold both traditional Christian convictions and a vision for how technology can serve human flourishing.

Christofuturist parents and funders are prime candidates to champion open-source adaptive AI tutoring. Why? They have strong values they want to preserve. They understand technology well enough to distrust closed systems. They have the resources and networks to fund alternatives. And they're motivated by something deeper than market returns—they want their children formed in truth.

If we're going to see the Linuxification of adaptive AI learning, it will likely be catalyzed by communities like this: people who are both technically sophisticated and spiritually serious, who understand that the infrastructure layer matters for long-term cultural formation.

The Church as Formation Hub: Youth-Led Renaissance

Here's a concrete vision for what Christofuturist schools could look like: church-based learning pods powered by open-source adaptive AI systems, where students don't just learn—they lead the revitalization of their parishes.

Imagine a network of pods, each with 10-15 students, hosted in church facilities. A soulful guide—perhaps a loving, motherly figure from the congregation—provides the human warmth, accountability, and spiritual formation. Meanwhile, adaptive AI tutors handle the knowledge drilling: math, reading, writing, whatever foundational skills each child needs, personalized to their level and pace.

But here's what makes this truly different: the curriculum is entrepreneurial and project-based, oriented around serving the church and local community.

Students aren't just learning in isolation—they're practicing leadership by helping their church thrive:

• Marketing and Communications: Students learn digital marketing by actually managing the church's social media, email newsletters, and community outreach. They build real skills while expanding the church's reach.

• Technology Systems: Students help modernize the church's CRM, donation systems, and member engagement tools. They're learning practical tech skills while solving real problems for their faith community.

• Youth Programs: Older students help design and run programs for younger children. They're learning leadership, teaching, and service while strengthening the intergenerational fabric of the church.

• Community Partnerships: Students reach out to local businesses, building relationships that benefit both the church and the community. They're learning sales, negotiation, and professional communication in a real-world context.

• Entrepreneurial Ventures: Students create small businesses—food sales, farmer's markets, service offerings—with proceeds supporting the church's mission. They experience the full cycle of identifying needs, creating value, and managing resources.

This model is self-reinforcing. The church provides physical space and spiritual grounding. The students provide energy, innovation, and labor that helps the church grow. As the church thrives, it can support more students. Success stories attract more families. The school becomes inseparable from the church's mission.

Why Churches Need This Partnership

Most churches face two persistent challenges: declining youth engagement and limited resources. A formation pod model addresses both simultaneously.

Young people don't want to be passive recipients of programming—they want to contribute meaningfully. When students are genuinely responsible for aspects of the church's operations and growth, they develop ownership and investment that passive attendance never creates.

Meanwhile, churches gain capable help with tasks that often go undone: updating the website, running social media, organizing events, managing databases, reaching new families. What church wouldn't benefit from a dozen motivated young people eager to serve?

The revenue model also makes sense. School choice funding (ESAs, tax credits, vouchers) provides roughly $8,000-10,000 per student in many states. A pod of 12 students generates meaningful revenue for the church while providing families with high-quality, values-aligned formation at accessible cost. Local businesses can sponsor students or programs in exchange for visibility through the church's network—everyone wins.

From Consumers to Contributors

Traditional schools treat students as consumers of education—passive recipients who absorb content and regurgitate it on tests. This model produces graduates who expect to be served rather than to serve.

The church-based pod model inverts this entirely. Students are contributors from day one. They're not preparing for some hypothetical future when they'll finally matter—they matter now, to their church, their community, their peers.

A student who spends years helping their church grow, serving their community, and practicing leadership will graduate with something no traditional transcript can capture: proven capacity to create value in the world. They'll have references from pastors, business owners, and community members. They'll have a portfolio of real projects. They'll have the confidence that comes from actually doing hard things that mattered.

And critically, they'll have a deep, lived understanding of what it means to serve Christ by serving others. Not abstract theology, but embodied practice.

The future of formation is:

• Soulful human mentors who love the children they're forming
• Open-source, modular AI systems that parents can customize to their values
• Church-based pods where students lead the renaissance of their parishes
• Project-based, entrepreneurial curriculum tied to real community impact
• Collective iteration that makes the whole system better over time

See: The Heroism Required for Fatherhood for understanding why parents need tools that work with their values, not against them.

A Call to Builders

Eventually, winning patterns emerge—meta-patterns. Humanity tends to converge on defaults. The question is: What are the good defaults for formation? The more those defaults point to open-source infrastructure that serves families rather than extracts value for shareholders, the better.

This is admittedly high-level. All the details need to be figured out. But I want to cast the vision because even if I don't know exactly how to make manifest progress toward this, I'm raising my hand as someone who would like to support—in small ways when I have time—people who want to contribute to this collective intelligence infrastructure.

The most beautiful thing would be humanity collectively building infrastructure that helps every individual kid have a very high quality education—or at least makes it a lot easier for many more kids to access high-quality formation.

Everyone wins when we contribute to shared, open-source, adaptive AI learning infrastructure.

If this vision resonates with you and you're building in this space, reach out—DM me on X at @garysheng.

---

"Train up a child in the way he should go; even when he is old he will not depart from it." — Proverbs 22:6

The way a child should go isn't determined by a billionaire's algorithm. It's determined by parents who know their children and the God who created them.