SkyWrite

SkyWrite is a browser-based educational tool designed to help children develop handwriting skills through air writing — using hand gestures detected by computer vision. The product is self-reported as an accessible learning companion that uses MediaPipe for gesture recognition, adaptive learning algorithms, and structured practice levels.

The project was built by one individual (Marie Rosario) in the context of a hackathon submission. It includes four progressive levels of scaffolding, supports 62 characters (uppercase letters, lowercase letters, digits), and is claimed to use objective performance measurements for progress tracking.

Key commercial due-diligence questions:

1. What is the actual educational impact or efficacy of SkyWrite?
2. How does the adaptive learning engine determine mastery?
3. Is there any evidence of user testing with target children?
4. What are the technical limitations around free-form writing?

The single most important open question: Is SkyWrite effective at improving handwriting skills, or is it merely a novel tech demo?

The description states that SkyWrite:

• Allows children to practice handwriting by writing letters or numbers in the air
• Uses MediaPipe to detect hand movements through a device's front-facing camera
• Compares the child’s movement with structured stroke paths
• Includes all 62 uppercase letters, lowercase letters, and digits
• Provides four progressively more challenging levels of practice:
  • Level 1: Follow the Trail (with shaded tracing corridor)
  • Level 2: Follow the Guide (no shaded corridor)
  • Level 3: Sky Writer (write from memory with hints)
  • Level 4: Free Write (write from memory with no guidance)
• Uses React, TypeScript, Vite, MediaPipe Tasks Vision and Hand Landmarker, HTML5 Canvas, Browser camera APIs, and Local browser storage
• Runs entirely in the browser without requiring local Python environments

Not evidenced:

• The actual mechanics of how stroke paths are compared or measured
• Whether the system tracks or evaluates specific motor skill development
• How the adaptive engine selects next characters or levels

The description states that SkyWrite is positioned as:

• An accessible web application for children with learning differences (autism, ADHD, dyslexia)
• A multi-sensory technique connecting letter recognition with large physical movements
• An educational tool built from a prior computer-vision experiment
• A way to transform an isolated ML prototype into something students can use

The project evolved from:

• A local Python prototype for two students
• A computer-vision experiment that was "not automatically a useful learning experience"
• Into a browser-based application with four distinct levels of scaffolding

Inferred:

• The author's intent is to make educational technology more accessible through web deployment
• There may be an emphasis on inclusive design for children with special needs

The description states that SkyWrite targets:

• Children who struggle with handwriting (especially those with autism, ADHD, or dyslexia)
• Students who find handwriting practice repetitive, frustrating, or difficult to remember
• Younger students who may not yet be able to read instructions
• Tutoring environments where the author spends 20 hours weekly

Not evidenced:

• Specific age ranges for target users
• Whether there are any formal partnerships with schools or educators
• The size of the potential market or user base
• Any demographic breakdowns of intended users beyond general learning differences

The description states that SkyWrite:

• Is a web application accessible through a link (no installation required)
• Does not appear to have any pricing information or monetization strategy
• Is described as a "free" tool for students to use
• Has no evidence of subscription, licensing, or transactional models

Not evidenced:

• Any revenue streams or business model
• Pricing tiers or customer acquisition costs
• Whether there are plans for monetization or commercial partnerships
• Any indication of B2B or B2C targeting

The description states that SkyWrite:

• Runs entirely in the browser using MediaPipe Tasks Vision and Hand Landmarker
• Uses React, TypeScript, Vite, HTML5 Canvas, Browser camera APIs, and Local browser storage
• Preserves educational structure of original tracing exercises
• Smooths coordinates to reduce camera jitter
• Maps hand movements onto a tracing canvas
• Checks whether the child is moving near the correct path and in the correct direction
• Divides each letter into intended strokes
• Requires students to begin at the correct starting point, follow the stroke, and fully reach its endpoint before advancing
• Uses one-second pauses between strokes
• Records performance metrics for adaptive learning engine

Inferred:

• The system is designed for low-latency interaction with real-time computer vision
• There may be limitations in accuracy or responsiveness due to browser-based processing

The description states that SkyWrite:

• Was built as a hackathon submission (OpenAI 2026)
• Has no evidence of revenue, customers, or adoption data
• Was initially an isolated ML experiment
• Was transformed from a local prototype into a web application
• Includes a data collection tool for Level 4 development

Not evidenced:

• Any user base or customer numbers
• Revenue figures or funding rounds
• Customer testimonials or usage metrics
• Product maturity beyond initial prototype stage
• Any evidence of market traction or adoption

The description states that SkyWrite:

• Is positioned as an educational tool for handwriting practice
• Uses computer vision and adaptive learning techniques
• Is claimed to be more accessible than traditional methods
• Was inspired by the author's previous work with MediaPipe
• Has no direct competitors mentioned in the description

Not evidenced:

• Any competitive landscape analysis
• Comparison to existing handwriting or educational technology tools
• Market positioning relative to other edtech products
• Evidence of similar technologies already in the market

The description indicates several potential risks:

1. Educational efficacy: The author notes that "a technically functional tracing exercise is not automatically a useful learning experience" — suggesting uncertainty about actual impact.
2. Technical limitations: Free-form writing (Level 4) has significant technical challenges, including difficulty distinguishing intentional from accidental marks.
3. Scalability concerns: The project was built by one person and lacks evidence of team expansion or institutional support.
4. Market validation: No evidence of user testing with target children or market demand.
5. Data quality issues: Difficulty finding suitable datasets for training free-writing recognition models.
6. Sustainability: No indication of long-term development plans, funding, or commercial viability.

1. What specific educational outcomes or improvements have been observed in children using SkyWrite?
2. How does the adaptive learning engine determine when a child has mastered a character?
3. Have you conducted any user testing with actual children, particularly those with learning differences?
4. What are the technical limitations of Level 4 (Free Write) and how do you plan to address them?
5. Is there a roadmap for monetization or commercial deployment?
6. How do you plan to scale beyond a single developer?
7. What is the current status of data collection for improving Level 4 accuracy?
8. Are there any partnerships with schools, therapists, or educational institutions?