Point of View
Education does not need more technology.
It needs better relationships between the technology it already has.
Education has accumulated platforms, portals, content libraries, assessment systems, communication tools and now AI interfaces. Many are useful. The deeper problem is that each one sees only a fragment of the learning process.
Educational intent lives in one place, activity in another and evidence somewhere else. People are left to reconstruct the relationships between them.
Our work begins with those relationships: between intent and execution, activity and evidence, evidence and action, people and AI.
01 · Featured thought
A chatbot is not an AI-native learning system.
Putting intelligence inside an interface does not make the surrounding system intelligent.
A chat window can produce an impressive response while remaining disconnected from the learner's objective, the educator's plan and the evidence already held elsewhere. The burden of supplying context—and judging whether the answer belongs in the learning process—falls back on the person typing.
AI-native architecture begins when learning goals, context, knowledge, authority, tools and evidence become available to intelligence as native parts of the system.
02 · Featured thought
Personalization changes what you see.
Adaptation changes what happens.
A personalized dashboard may know your name, language and preferences. It may reorder content or recommend something similar to what you selected before.
An adaptive system carries a stronger responsibility. It understands enough about the journey to change the next step based on purpose, evidence and context—and to explain why that change is appropriate.
Sometimes adaptation means a different resource. Sometimes it means more time, a new route, a human conversation or no change at all.
03 · Featured thought
The LMS is not disappearing.
It is becoming one service among many.
The administrative functions of the LMS remain useful. Organizations still need enrollment, access, delivery and reporting. The mistake is asking one application to remain the boundary of digital learning when learning already moves through many environments.
The next learning system will extend across services and experiences. The LMS can remain part of it without being asked to represent the whole.
04 · Featured thought
The model is replaceable.
The learning system is not.
AI models will evolve quickly. Different models will be appropriate for different roles, languages, costs and levels of risk. Organizations should be free to make those choices without rebuilding their learning architecture every time intelligence improves.
Context, institutional knowledge, permissions, learning structure and evidence belong to the organization. They are the durable part of the system and the source of continuity when technology changes.
Models should plug into that world. Not own it.
Continue the conversation
What should education infrastructure make possible?
Our point of view is not a feature list. It is an argument for a learning system that connects purpose, participation, evidence and action.
These ideas are open questions as much as positions: how should human authority be represented in an agentic system? What context should follow a learner? When should evidence trigger action? Which parts of institutional memory must remain independent of any model or vendor?
Those are the conversations from which better architecture begins.