The Problem
AI that understands
how you
learn.
Dhilabs is an applied AI research company building adaptive learning systems that understand what a person knows, where their understanding breaks down, and what will help them progress next.
It is understanding what each person needs next.
AI can retrieve information, generate explanations and answer almost any question in seconds. But access to an answer is not the same as learning. Two people can ask exactly the same question while bringing different knowledge, misconceptions, goals, confidence, memory and context. The next challenge is not giving everyone more information.
A Brief History of Learning
Every technology changed
access to knowledge.
THE TUTOR
Learning began as something deeply personal. One teacher. One learner. A tutor could notice confusion, change an explanation, revisit an idea or move ahead. Highly adaptive. Difficult to scale.
THE CLASSROOM
Education became scalable. One teacher could reach dozens of learners at once. Education expanded dramatically, but adaptation became harder. Different learners increasingly moved through the same material, sequence and pace.
THE INTERNET
Knowledge became abundant. Search, online courses, video and digital libraries put extraordinary information within reach. But the learner still had to decide what to learn, what to trust and what to do next.
THE AI ASSISTANT
Answers became abundant. AI can explain almost anything instantly, but may not know why the learner asked, what they understand, misunderstood, forgotten or are ready for next.
THE ADAPTIVE SYSTEM
Learning can become personal again. Software can observe how someone learns over time, model their understanding and continuously adapt the experience around them. This is the transition DHI is working on.
The Dhilabs Thesis
The next breakthrough in learning isn’t more information.
It’s better understanding.
We believe the next generation of learning systems will do more than generate content or answer questions. They will maintain an evolving understanding of the learner, including their knowledge, misconceptions, goals, context, memory, confidence and progress, and use it to decide what should happen next. Not simply: What is the answer? But: What does this person need now?
The Learner Model
What does it mean for AI to understand a learner?
A learner model is a continuously evolving representation of what a person knows, where their understanding is incomplete, how they respond to different learning experiences and what may help them progress next. No single signal is enough. Select one below to explore it.
MODEL UPDATE / 01
Knowledge
What is already understood?
Skips what you know. Starts where you stopped.The model is never finished. Every interaction can update it.
The Dhilabs adaptive loop
Observe. Understand. Adapt.
An intelligent learning system shouldn’t simply respond.
It should learn from every interaction.
Then it happens again.
The learner changes after every experience, so the system updates its understanding and adapts again.
Beyond Personalization
Personalization changes what you see.Understanding changes what happens next.
Most personalization begins with preferences: which content should this person see, which format do they prefer, and which recommendation is most relevant? We are interested in something deeper. A learning system should reason about the learner’s current state and adapt because of what it understands.
- 01 / Personalization
- “Show this person something different.”
- 02 / Learner understanding
- “This person understands A, is uncertain about B, has a misconception about C and is ready for D.”
Applied Intelligence
We turn these ideas into real learning systems.
This is where our thesis meets a real learner.
Outlyn
Adaptive AI learning for GATE CSOutlyn understands where a learner stands, builds a living preparation plan and adapts what they should learn, practise or revise next. We’re starting with GATE Computer Science and working to expand to other GATE exams.
Explore Outlyn →Human Development
Can intelligent systems help people build complex capabilities over months and years?
Adaptive Intelligence
What happens when software maintains an evolving model of the person using it?
Research
Questions we’re
trying to answer.
These questions shape what we build, test and learn next.
Can AI accurately model what a learner understands?
Can AI detect misconceptions before the learner recognizes them?
How should a learning system decide what someone should learn next?
How should memory and forgetting change the learning experience?
Can learning become continuously adaptive instead of curriculum-bound?
What happens when software remembers how you learn?
Thinking
Ideas about AI, learning and human intelligence.
These working notes show the ideas behind the systems we are building. Longer-form thinking will be published when it is ready.
Why knowing isn't understanding
Why access to information is only one part of learning.
Read the researchThe learner model
What should an intelligent learning system understand about the person it is helping?
Read the researchBeyond personalization
Why selecting different content is not the same as understanding a learner.
Read the researchWhen mistakes become data
What errors, hesitation and misconceptions can reveal about understanding.
Read the researchThe adaptive system
What changes when software continuously adapts around an individual?
Read the researchOur Manifesto
We believe technology should adapt to humans.
Not the other way around.
We believe learning is personal.
We believe access to information is not the same as understanding.
We believe mistakes contain information.
We believe misconceptions matter.
We believe memory changes what someone needs next.
We believe curiosity is a signal.
We believe understanding matters more than recall.
We believe the learner should shape the system.
And we believe intelligent software can help make that possible at a scale that wasn’t possible before.
This is what we’re building.
Team
Meet the people building DHI.
Our founding team brings together company building, machine-learning research and product. Together, we are exploring how intelligent systems can understand people and adapt to them over time.
Meet the team →Careers
Build things that shouldn’t be possible yet.
We’re interested in researchers, engineers, designers and unconventional thinkers fascinated by intelligence, learning, cognition and adaptive systems. If you care about understanding how people learn and building technology that can respond intelligently, we’d like to hear from you.
Explore careers →