KEEP by STUDIO B

The company appears to be a solo project (1 person) submitted to the OpenAI 2026 hackathon, named KEEP by STUDIO B. The author states it is a web application that helps creators build structured, modular prompts for AI image editing, aiming to improve consistency and reduce unintended changes in generated images. It uses GPT-5.6 and Codex for prompt generation and is built with Next.js, React, Tailwind, TypeScript, and Vercel.

What changed: The project evolved from a monolithic prompt approach to modular blocks (KEEP, CHANGE, QUALITY) based on testing insights. This shift reflects an understanding that structured prompts may be more reliable than long ones.

The single most important open question: Is there any evidence of user adoption or feedback beyond the hackathon submission? The description lacks any data on usage, customers, revenue, or product-market fit beyond self-reported claims.

• The description states that KEEP is a web application.
• It is built using Codex, GPT-5.6, Next.js, React, Tailwind, TypeScript, and Vercel.
• It helps creators build structured, modular prompt blocks (KEEP, CHANGE, QUALITY) for AI image editing.
• It takes a reference image as input and uses GPT-5.6 to convert it into character specifications.
• Users can then copy only the sections they need, making prompts easier to edit, reuse, and adapt.

Note: The product is described as a prototype or hackathon submission. No evidence of commercial deployment, user base, or monetization exists in the description.

• The author states that AI image editing often changes unintended elements like faces, identity, clothing, or style.
• The solution is positioned to help creators communicate exactly what should stay the same, aiming for character consistency.
• The project evolved from a monolithic prompt approach to a modular one, based on testing insights.
• The author claims that prompt structure matters as much as content, and that modularity improves usability and consistency.

Inference: The positioning reflects an understanding of user pain points in AI image generation, but no evidence of market validation or customer feedback is provided.

• The description states that KEEP helps creators who use AI image editing.
• It targets users who want to preserve character identity and style during edits.
• It is aimed at those who may struggle with complex or unreliable prompts, especially in AI image generation workflows.

Not evidenced: No specific customer personas, user segments, or buyer profiles are described. The ICP is inferred from the stated use case but not defined.

• Not evidenced.
• There is no mention of pricing, monetization strategy, or business model in the description.
• The project is presented as a hackathon submission with no indication of commercial intent or revenue streams.

• Built with Next.js, React, Tailwind, TypeScript, and Vercel.
• Uses Codex and GPT-5.6 for prompt generation and image analysis.
• The app is a web application.
• The workflow involves:
  • Taking a reference image
  • Converting it into structured character specifications using GPT-5.6
  • Generating modular prompt blocks (KEEP, CHANGE, QUALITY)
  • Allowing users to copy only the needed sections

Inference: The technical stack suggests a modern frontend with backend integration for AI processing. However, no evidence of scalability, performance, or infrastructure details is provided.

• The project was submitted to the OpenAI 2026 hackathon.
• It is described as a prototype, not a production product.
• No evidence of:
  • Users
  • Customers
  • Revenue
  • Product-market fit
  • Adoption metrics
  • Growth or retention data

Absence of evidence: There are no signs of traction, usage, or market validation beyond the hackathon submission.

• Not evidenced.
• No mention of competitors, existing solutions, or market positioning is provided in the description.
• The author does not reference similar tools or platforms for AI prompt engineering or image editing.

• Solo team: Only one member (Soregashi STUDIO B) is listed. This raises questions about execution capacity and scalability.
• Hackathon prototype: The project is a hackathon submission, not a commercial product — no evidence of real-world use or development beyond the event.
• No monetization strategy: No indication of how the tool will be monetized or whether it has a path to revenue.
• Unverified claims: All features and benefits are self-reported without external validation.
• No customer feedback or usage data: The lack of any traction signals raises concerns about product-market fit.

1. What was the actual user feedback during the hackathon? Did anyone use it beyond the event?
2. Are there any plans to expand beyond the current prompt modules (KEEP, CHANGE, QUALITY)?
3. How does the tool handle different AI image models or APIs? Is it model-agnostic?
4. What is the long-term vision for monetization or product development?
5. Has the team tested the tool with real users outside of the hackathon?