LIESSON

The company appears to be a single-person educational tech project named LIESSON, which uses AI to teach by intentionally embedding errors into learning content. The app invites users to identify false statements within lessons and explains why certain answers are incorrect. It is built using GPT-5.6, Next.js, Node.js, OpenAI API, Playwright, PWA, React, Tailwind CSS, TypeScript, Vercel, Vercel KV, Vitest, Zod, and other tools.

The project is described as a hackathon submission to the OpenAI 2026 hackathon. It has no demonstrated traction, revenue, or customer base. The author states that the app uses AI to generate lessons with deliberate errors based on a five-tier scale of subtlety, and that it includes features like encrypted answer keys and rate limiting for API use.

The single most important open question is whether this concept can be scaled into a viable product with sufficient educational value and user engagement.

• The description states that LIESSON is an education app that teaches by lying to users.
• Every question in the app contains 1–3 deliberate errors.
• Users are invited to identify these errors and explain why they are incorrect.
• The app incorporates learning-science research showing that error detection improves encoding compared to studying correct material.
• It allows users to practice identifying false statements within lessons.
• The author claims GPT-5.6 generates the lesson content, determines error subtlety (on a five-tier scale), and grades user explanations.
• The app uses Codex for engineering, with API routes built using strict JSON schemas.
• Session data is encrypted server-side to prevent exposure in the browser.
• Teacher-upload functionality allows PDF/text extraction entirely in the browser without touching the server.

This is a self-reported description of an educational tool that leverages AI-generated misinformation as a pedagogical technique. No evidence of actual product usage, customer feedback or revenue exists.

• The author states that no consumer app has ever made "the lie itself the core mechanic."
• The app positions itself around a learning-science principle: using erroneous examples to promote deeper encoding.
• It claims to bring an “underrated side of learning” into the AI space.
• The project is framed as a novel approach to education, not just another learning platform.
• There is no indication of prior positioning or evolution in messaging beyond this single description.

This is a claim about uniqueness and educational innovation, but there is no evidence of market validation or prior iterations of the idea.

• The description does not identify specific target customers or personas.
• It implies that anyone who wants to learn or practice could use the app.
• The author mentions potential partnerships with schools, exam boards, tutors — suggesting an institutional or academic audience.
• No explicit segmentation or targeting beyond general learners is evident.

Not evidenced. The description lacks any customer profiling or ICP definition.

• The description does not mention a business model or pricing structure.
• There is no indication of monetization plans, subscription tiers, or revenue streams.
• The author notes that the app uses two model calls per lesson and that rate limiting was implemented to avoid API key depletion — but this does not imply a paid service.

Not evidenced. No commercial framework is described.

• Built with: Codex, GPT-5.6, Next.js, Node.js, OpenAI API, Playwright, PWA, React, Tailwind CSS, TypeScript, Vercel, Vercel KV, Vitest, Zod.
• The app uses structured outputs and strict JSON schemas for API routes.
• Session data is encrypted server-side to avoid exposure in the browser.
• Teacher-upload feature extracts PDF/text content entirely in the browser.
• Rate limiting implemented using Vercel KV to prevent API key drain.
• The author used a single long CLI session with Codex to build most of the app.

These technical choices suggest a developer-focused, AI-integrated prototype. No evidence of production-grade infrastructure or deployment beyond a demo is provided.

• The project is described as a hackathon submission.
• It has no demonstrated traction, users, or adoption metrics.
• The author states that the app was built in one session using Codex and includes only one team member (Rafael Marc).
• No mention of user engagement, retention, or usage statistics.

Not evidenced. No signs of product maturity or traction are present.

• The description does not reference existing competitors.
• It claims no consumer app has ever used lies as the core mechanic in education.
• There is no indication of awareness of similar tools or platforms in the educational technology space.

Not evidenced. No competitive landscape is described.

• Unproven pedagogical efficacy: The approach relies on a learning-science principle, but there is no evidence that this method works better than traditional methods.
• Single-person development: The app is built by one person, which raises questions about scalability and long-term maintenance.
• AI dependency: Heavy reliance on GPT-5.6 and OpenAI APIs may create cost and availability risks.
• Lack of commercial viability: No business model or monetization strategy is described.
• Potential for misinformation misuse: The app’s premise of embedding lies could be misused in contexts beyond education.

These are inferred risks based on the self-reported description, not verified facts.

1. What evidence supports the claim that error-based learning improves encoding more than correct material?
2. How does the app ensure accuracy in non-error content when it uses AI to generate lessons?
3. Is there any user testing or feedback on how effective this method is for learners?
4. What are the plans for monetization and scaling beyond a hackathon prototype?
5. How will the app handle potential misuse of its "lying" mechanism outside of educational contexts?
6. What is the long-term vision for the product, and how does it intend to evolve from a demo into a full product?