The problem Auditsseus solves

1. It is difficult to detect plagiarism issues in AI-generated content, especially style plagiarism
2. Many similar IP issues also occurred in early protocol-related white papers, theses, etc.
3. IP registration and verification can be verified through manual direct search, making it difficult to protect in advance

Solution

IP audit and plagiarism detection:

• Vectorize uploaded content and compare it with existing IPs in real time
• Calculate semantic similarity through Eliza + Story Plugin
• Supports various formats including text format through document parsing via Upstage API and image file format.

AI-based 'Style plagiarism' detection:

• Learned AI determines the style, color, and structural similarity of images
• Continuously improves the model through user feedback and actual upload data

Automatic on-chain registration:

• Automatically registers IPs in Story Protocol and registers IPs in reports when users select them based on plagiarism rate reports

Overview

Auditsseus is an AI-based IP audit secretary service for preventing plagiarism of AI-created works and protecting IPs. A Web3 creative work protection platform that analyzes plagiarism rates by performing vector comparisons with Story Protocol's IP database for user-uploaded content (images, papers, PDF, etc.) and assists with IP registration through the Eliza agent.

"In an era overflowing with AI creative works, we are creating an AI secretary and audit system that protects the legitimate rights of IP."

Challenges we ran into

About Upstage:

• firstable, there's an issue that using upstage api. the primary one is there's a limitation in api key. even i didn't used all of credits. so we have to wait for sometimes. and when i build elizaos plugin-upstage, in document, they used FormData, that is deprecated one. and in eliza, they using another FormData that exactly same spell. so it has occurred error. so i spent much time to solve that.

About Story:

• While using the Story Protocol plugin, we faced several challenges. Since the plugin was community-built and not actively maintained, many functions were deprecated, making integration difficult. The SDK was initially based on the Odyssey network, and although it was partially updated during the hackathon, some inconsistencies remained. Testing IP registration was also challenging. The simple IP Portal only supports mainnet, so registering test IPs required using the SDK or interacting directly with contracts. Having a dedicated testnet IP Portal would greatly improve the development experience. Registering IPs via StoryScan was complex, with unclear return values and a lack of intuitive feedback. We found the IP registration process overly complicated and plan to contribute a simplified version of the Eliza plugin via pull request. More detailed documentation on the registration process, required data formats, and complete API references would be very helpful.

Challenges:

• Frequent dependency issues between deployed Story Protocol contracts and AI agent interactions.
• No batch comparison support—each collection had to be queried individually.
• For production use, backend infrastructure will be necessary to support scalable comparison beyond single images. (To solve this, we will plan to utilize Eigenlayer DA.)

Learning:

• We learned how to extend the story-plugin by adding custom actions.