Socratic AI Tutor

A Self-Regulated Learning App for AI-Assisted Problem Solving

 Open App  See Full Interventions Project

Note: The app requires activation before use. If you'd like to try it, please email me and I'll get it running for you.

The Socratic AI Tutor is a Streamlit web application built to address a core problem identified in classroom AI research: when students are given free access to AI tools, they typically use them to bypass thinking rather than scaffold it. The app guides students through a structured, step-by-step workflow rooted in Self-Regulated Learning (SRL) theory, generating ready-to-paste prompts that instruct AI to act as a Socratic tutor — asking guiding questions rather than giving direct answers.

6
Guided Steps
3
Download Formats
Works with Any Subject
Works with ChatGPT & Claude

The App

Socratic AI Tutor App Screenshot

Open App

Note: The app requires activation before use. If you'd like to try it, please email me and I'll get it running for you.

Why This Approach

Self-Regulated Learning & the Socratic Method

Self-Regulated Learning theory holds that durable learning requires students to actively monitor their own understanding, set goals, and adapt strategies — rather than passively receiving information. The Socratic method, in which a teacher guides a student toward understanding through questions rather than answers, is one of the oldest pedagogical implementations of this principle.

The problem with unstructured AI use is that it short-circuits both: students outsource the thinking and receive answers without developing the metacognitive process that makes learning stick. This app redirects AI into the role of a Socratic scaffold, keeping the student in the driver's seat at every step.

The central question embedded throughout the workflow: "Is this helping me achieve the learning objective, or is this beating the assignment?"

Technology Acceptance Model (TAM)

Davis (1989) showed that technology adoption is driven by perceived usefulness and ease of use. Albayati (2024) applied TAM to ChatGPT, finding that trust, privacy concerns, and social influence all shape student attitudes toward AI — independent of whether the tool actually helps them learn. The Socratic Tutor app was designed with both dimensions in mind: it is low-friction to use (no login, works on any device) while being explicitly structured to increase usefulness for actual learning rather than just task completion.

How the App Works

  1. Student Setup Student enters their grade level and course (e.g., "11th Grade, AP Microeconomics"). This context is embedded in every subsequent AI prompt automatically.
  2. Concept Review Before touching the problem, the student lists the concepts they think they'll need. This activates prior knowledge and makes gaps visible before the AI interaction begins.
  3. Problem Input Student pastes the specific problem. The app generates a complete Socratic priming message to copy into ChatGPT or Claude — instructing the AI not to give the final answer and to scaffold learning step by step.
  4. First Step Check Student indicates whether they know the first step (yes or no). The app generates a tailored follow-up prompt for each path, maintaining the Socratic structure regardless of starting confidence.
  5. Outcome Reflection After working with the AI, the student returns and logs their outcome: SUCCESS (request a similar practice problem), RETRY (request a rephrased explanation), or MASTERED (confirm completion). Each generates a corresponding follow-up AI prompt.
  6. Session Download The app produces a downloadable session record in three formats: a full markdown transcript, a concise summary, and a JSON file formatted for Toddle submission. This creates an accountability record and an artifact for teacher review.

Technical Details

Built in Python using Streamlit and deployed via Streamlit Community Cloud. No database or authentication required — session data is handled in-app and exported by the student at the end.

Python Streamlit Pandas JSON / CSV export Streamlit Community Cloud UUID session tracking

Design Decisions

  • No login required — removes friction for classroom use and protects student privacy
  • Works with any AI tool — generates copy-paste prompts rather than calling an API directly, so students use their own ChatGPT or Claude accounts
  • Three export formats — markdown transcript for teacher review, simple notes for student reference, JSON for Toddle integration
  • Branching logic — the yes/no first-step check and success/retry/mastered reflection produce different follow-up prompts, personalizing the workflow without requiring a complex backend
View on GitHub