Create anywhere. 
Verify in Drill.

Keep using the AI workflow you trust. ChatGPT, Claude, PDFs, notes, and lectures can all produce useful study material.

Drill starts after generation. It turns draft material into inspectable Cards, tests it through Truths and Traps, and tracks whether knowledge is Untested, Weak, Unproven, Mastered, or Due.

Workflow

AI is not the verdict.

Generated content can be useful and still be wrong. It can sound complete and still miss the point. It can produce questions without proving that you know the answer.

Drill does not ask AI to decide what is true. It gives your material a verification protocol.

01.

Generated content can be useful and still be wrong.

It can sound complete and still miss the point. It can produce questions without proving that you know the answer.

02.

Drill does not ask AI to decide what is true.

It gives your material a verification protocol.

03.

Keep the source in charge.

The student reviews the source, owns the truth, and decides what can become a Card, Truth, or Trap.

04.

End in the Drill protocol.

Cards hold the material. Truths and Traps test it. Runs expose misses. Readiness shows what has evidence.

Draft

Draft Cards from real exam material.

A useful AI draft starts with source material: PDFs, notes, slides, past papers. Ask for small Cards and testable Statements. Then inspect before you study.

Prompt 01Lecture PDF

Turn a PDF into Cards shaped for verification, not a summary that feels familiar.

Turn this lecture PDF into atomic exam Cards. Each Card should test one idea.
Prompt 02Class notes

Turn notes into Truths and plausible Traps that expose the confusion an exam can punish.

From these notes, write true Statements and plausible false Traps that would expose confusion.
Prompt 03Past paper

Use past papers to find concepts to drill, not questions to memorize by shape.

Use this past paper to identify concepts I need to drill, not just questions I can memorize.
Inspect

Inspect the Truths and Traps before studying.

AI flashcard generators optimize for fast decks. Drill treats the deck as unfinished until you can see the claims, edit the Traps, and remove weak material.

  • Split bundled ideas

    A generated flashcard often hides three claims inside one prompt. Split it until one wrong answer points to one gap.

  • Check standalone Statements

    A Truth or Trap should make sense without the PDF open beside it. If the Statement depends on context, rewrite it.

  • Make Traps plausible

    Bad AI decks contain false answers no student would choose. A Trap should pivot one detail and reveal real confusion.

  • Delete filler

    More Cards do not mean more preparation. Remove definitions, summaries, and trivia that will not matter under exam pressure.

  • Keep the source nearby

    When a Card looks suspicious, compare it to the lecture, slide, note, or past paper before it enters your Run.

  • Approve before testing

    AI output is material. Drill makes it visible and editable before it becomes something you measure.

Verify

Your AI creates. Drill verifies.

AI can help draft study material from notes, PDFs, lectures, and past papers. That does not make the material true, complete, or exam-ready.

Drill treats AI output as draft material. The student reviews the source, owns the truth, and uses Drill for the verification loop: Cards, Truths, Traps, Runs, and Readiness.

Untested is visible.

Generated Cards do not become proof because they exist. They stay visible as material that still has to survive a Run.

Weak stays exposed.

A miss is not buried under a score. Weak Cards keep their state until recall holds under another pass.

Readiness is earned.

Readiness is not the number of Cards an AI produced. It is the material that has been tested, corrected, and proven through Runs.

Your AI can create the first draft. Drill shows whether you actually know it.

Technical setup

Connect the conversation when your AI supports it.

You can always copy useful output from ChatGPT, Claude, or another AI into Drill. When the tool supports connectors, the workflow gets tighter.

MCP stands for Model Context Protocol. It lets an AI tool work with apps like Drill from inside the conversation, so a draft can become Cards, Truths, and Traps with less copy-pasting.

Use ChatGPT where you already ask for the draft. Use Claude when you want the connector workflow today. Either way, the AI creates and you approve what enters Drill.

The protocol is setup detail. The value is simpler: source material becomes Cards you can test.

Setup

Technical setup after the material is clear.

ChatGPT is an app flow. Claude is a connector flow. Same rule at the end: inspect before the Cards become study material.

Step 01

Find Drill in ChatGPT

Open Drill from the direct ChatGPT app link. You can also find it from Apps in the sidebar, or Settings -> Apps & Connectors.

Step 02

Connect your Drill account

Click Connect. Sign in to Drill, review the access request, and authorize the connection.

Step 03

Add Drill to the conversation

Open a chat and add Drill from the app picker, or mention it inline. Then ask for the work: 'Create Cards from these lecture notes' or 'Turn this PDF into Traps and Truths.'

Open Drill in ChatGPT

Available through ChatGPT Apps. Some workspaces may require app access to be enabled by an admin.

Read the full ChatGPT Apps documentation
Verify the draft

Use AI to create study material. Use Drill to prove it holds.

Keep ChatGPT, Claude, and the habits that already work. Bring the useful output into Drill, inspect the Truths and Traps, then Run it.