Scratchpad: where kids write math — stories, models, proofs

The description states that Scratchpad is a free, no-login page where students arrange equations, tape models, ten-frames, number lines, long division, proofs, two-way tables, story text, art, and page breaks for long illustrated math stories. Every part remains machine-readable.

It also states that the tool connects to GPT-5.6 Sol to provide feedback on student work, including:

• Step-by-step analysis
• Misconception identification ("added the denominators")
• Feedback tied to exact steps
• Support for both story and calculation-based inputs
• Ability to generate prompts based on grade band and standards

The tool is built using technologies such as React, Node.js, Supabase, Vercel, TailwindCSS, TypeScript, OpenAI API, and Codex.

Inference The product appears to be a web-based math authoring and feedback platform designed for K–12 education. It is not a commercial SaaS offering but rather an open-source or prototype tool built for personal use by the founder.

The author claims that Scratchpad addresses two core problems in math education:

1. Current autograders only check answers, not thinking.
2. Worksheets bore students who enjoy math most.

It positions itself as a way to let students "tell stories with math" and provide useful feedback on both the story and its math together.

The evolution described shows a shift from a basic authoring tool to one that incorporates AI for checking work and providing coaching feedback.

The description states that the primary users are:

• Students in K–12 classrooms
• Teachers who want to give detailed feedback on student work
• Tutors or study partners who support learning

It also mentions that the tool supports "every part remains machine-readable" and allows students to write in their own language, including Spanish.

Inference The ICP likely includes educators teaching math in elementary through middle school grades, particularly those interested in integrating storytelling into math instruction.

The description explicitly states that Scratchpad is:

• Free
• No login required
• Does not collect student data

There is no mention of any pricing model or monetization strategy. The author notes that funding through prizes, OpenAI credit support, or donations would allow raising file-reading limits while keeping the tool free and login-free.

The project uses:

• GPT-5.6 Sol for feedback generation
• Codex for integration and planning
• React, Node.js, Supabase, Vercel, TailwindCSS, TypeScript, OpenAI API
• Serverless architecture
• LaTeX/KaTeX rendering
• Stable IDs for steps to ensure accurate referencing

It supports:

• Machine-readable components (equations, models)
• Feedback tied to specific steps
• Prompt generation based on grade level and standards
• File reading with rate limits
• Mobile layout considerations
• Accessibility features

Inference The technical stack suggests a lightweight, serverless web application built for educational use. It is not described as scalable or enterprise-grade.

The description does not provide any evidence of:

• Revenue
• Customers
• Adoption metrics
• Usage data
• User base size
• Product maturity beyond prototype stage

It mentions that the author tested "twenty different student pages across four grade bands, all forty-nine Scratchpad formats, and seven complete product flows."

However, this is not sufficient to indicate traction or market validation.

The description does not mention any competitors. It focuses on solving a gap in current autograders and worksheet-based instruction.

Inference The competitive landscape is unclear, but it likely overlaps with tools for math education, AI-powered grading systems, or storytelling platforms for learning.

• No revenue model or monetization plan: The tool is free and lacks commercial viability.
• Single founder: Only one team member (Jason Mortensen) is listed.
• Self-funded feature: The AI integration relies on self-funding, which may limit scalability.
• Limited testing scope: Testing was done within the author’s classroom context only.
• Unverified claims: All statements are self-reported and unverified.

1. What is the actual classroom usage beyond your own teaching?
2. How many students have used the tool outside of your class?
3. Are there plans to monetize or scale the product beyond its current prototype stage?
4. Has the AI feedback been validated by other educators or researchers?
5. What are the technical limitations of the current implementation that could prevent scaling?