Learning tools

Autodidact

A place to study with Faraday, an AI tutor: choose a subject, work through it, and return to what needs practice.

LearningRustReact

I’m building Autodidact to help people learn independently.

You bring a subject, your materials, and the time you have. Faraday, the AI tutor, helps you choose where to begin. You read, work through a problem, and talk through what you don’t understand. When you stop, your work is saved. When you return, there is a place to start.

Suppose you want to understand neural networks. There is plenty to read. Faraday helps you choose a passage, work through an example, and explain why it works. You can ask for help, use voice, or explore the rest of the course.

The harder question is what you learned. Following an explanation can feel much like understanding it. Trying it alone tells you more. Autodidact keeps track of where you needed help and brings the material back for practice.

I lead design and engineering for Autodidact.

Visit project ↗

How it works / Explore the workflow

From today’s study to the next session

The plan chooses a piece of the course to work through. The reader and Faraday stay with that step. Autodidact saves what you did, including where you needed help, and schedules later practice. It checks consent and course access before sending your text to a model.

Follow the left column from Begin through the current activity to the saved learning history and the scheduler, which brings material back for later practice. Orange boxes are checks. Faraday and the examiner sit in the middle column; outside services are on the right. Open full size ↗

Choose what to study today and leave yourself somewhere to begin tomorrow.

Choose a subject and bring any materials you want to use. Import a course or ask for one based on your goal, then look through what it covers.
Goal + material + route
Choose a piece of the course with a clear objective and a place to stop. The rest remains available in Explore.
Today’s objective + reading portion
Explain an idea, recall what you read, or work through an example. Faraday helps with the step you are on, and the session keeps track of that help.
Practice + feedback + help used
Save what you worked through, how it was checked, and where you needed help. An explanation reached with a hint and one given alone tell us different things.
Work saved + understanding checked
Leave with your place saved and a next step. Earlier material comes back when it is time to practice again.
Saved place + next practice

What happens when…

A hint helps you finish

Faraday gives you a hint and you finish the explanation. The session keeps track of the help. Trying the idea again on your own will show what you can do without it.

With a hint → try it on your own

Learnings

When you study on your own, deciding what to work on can take up much of the time you have. A course outline helps you see what there is to learn, but you still need to choose what to do today. If you have ten minutes, the software should help you pick a useful exercise or a short passage you can work through in that time.

Building Autodidact has made me pay more attention to those choices. How much reading does an exercise require? What should you be able to explain afterward? Where can you stop and return later? These questions shape the study session as much as the subject does.

There is also a difference between following an explanation and solving a problem yourself. A tutor can help you through a difficult step, but finishing with help does not tell you whether you could do it again alone. Keeping track of that help lets the software suggest what to practice next.

Explore the other projects →