Three years ago, I was tinkering with early prototypes of Mindsmith. As a college student, my first idea was to build a system that could elevate open-source textbooks. Millions of students use large repositories of open-source textbooks, but they still represent a relatively low form of learning: static text, passively consumed.

I built a complicated system to ingest the textbooks, but when it came down to it, the AI wasn't capable of genuinely improving them. I had seen glimpses of what a higher form of learning could look like in Duolingo, Quizlet, and others like them: interactive, adaptive, engaging. But none of them were general enough to teach something like a full college physics course.

From there, I realized the gaps were twofold:

1. There was no end state for learning that fully pushed the limits of what was possible.

2. AI was not capable of producing that hypothetical higher form of learning.

The years since can be summarized simply: building a medium worthy of the world's learning, and building the system to let instructional designers create it.

We had to be quite clever. We started as a micro-learning tool with simple text and image cards. We expanded to a full page layout system, and we currently have interactive, branching activities in the form of our Experiences builder. We've had this vision for Experiences to be micro-simulations but called them only "Scenarios" in the app because the technology wasn't there to generate true simulations.

The original experiment focused on college curriculum, but we found our path to this vision by focusing on learning teams inside large organizations: the instructional designers, subject matter experts, and training professionals who produce learning at scale every day. By building for them, we built the system that now makes the original vision possible.

What I'm seeing now inside our Mindsmith agent, an AI collaborator that works alongside instructional designers to author learning from start to finish, are glimpses of a system that could finally do what I was trying to do in college: elevate learning en masse at high efficiency.

One constraint that has been lifted is the idea of having to do it without any human interaction. When we partner with our customers and their instructional designers, they bring their insights and knowledge to the creation process. The agent brings speed, consistency, and creative range. As this collaboration matures and truly lets our customers realize their vision, one could imagine entire curricula and entire training programs rewritten in a matter of months.

Another constraint that has fallen away is the difficulty of creating content. The types of e-learning inside Mindsmith were previously constrained by what the AI could produce. Our Experience builder was called "Scenarios" because that's all we could reliably generate. Our video builder was designed to be elegant and minimalist, an alternative to text rather than full-fledged video editing.

Mindsmith was a patchwork of AI systems accelerating workflows, but with this new generation of AI models, we can generalize our system and give it the capabilities to produce any form of e-learning our platform supports, in the author's exact workflow.

You'll see this in our releases over the next few months. We've had prototypes of software simulations based on screen recordings, engaging story-style learning with consistent characters, and conversational practice for real-world skills. All of these can be seamlessly handed to our agent, with the agent capable of following your exact instructions.

We're excited about the future of e-learning, and we're putting our money where our mouth is. Over the next 30 days, we will increase headcount by 40%, we expect to more than double the team over the next year.

But we need your help. With your vision for learning and your scrutiny of the content, instructional designers working with our agent become something more: e-learning artisans, capable of elevating the world's learning at a pace and quality never before possible.