Inspiration
Indonesia has over 12 million content creators and a $38.5 billion creator economy — yet there are zero affordable IP protection tools available in Bahasa Indonesia. We kept seeing the same story: an artist posts original work on Instagram, and weeks later their design shows up on Tokopedia or Shopee slapped onto phone cases, t-shirts, or stickers — with no credit, no permission, and no payment.
The worst part isn't the theft. It's that most creators don't know they already have rights under Indonesian law. UU Hak Cipta No. 28/2014 grants automatic copyright protection the moment a work is created. But the law is written in dense legal Indonesian, enforcement information is scattered, and hiring a lawyer costs more than most creators earn in a month.
We built MineMap to close that gap — giving Indonesian creators a practical, AI-powered tool to understand their rights and actually do something when those rights are violated.
What it does
MineMap is an AI-powered IP rights platform built specifically for Indonesian creators. It walks users through the full journey from "someone stole my work" to "here's what I can do about it."
- Legal Q&A in plain language — Ask questions about your IP rights and get answers grounded in Indonesian law (UU Hak Cipta 28/2014, UU Merek), in Bahasa Indonesia or English
- Infringement assessment — Describe your situation through a structured intake form and receive a qualified legal analysis of whether infringement likely occurred
- Takedown request generator — Produce pre-filled takedown requests for Tokopedia, Shopee, YouTube, Instagram, and TikTok, mapped to each platform's actual form fields
- Evidence tracker — An interactive checklist that guides creators through gathering the proof they need (screenshots, timestamps, original files) and persists across chat sessions
- Printable case report — A formatted report with auto-extracted law references, suitable for filing complaints with platforms or DJKI (Indonesia's IP office)
How we built it
- Frontend: Next.js 14 with TypeScript, styled with Tailwind CSS and DaisyUI (autumn theme) for a warm, approachable interface
- AI backbone: DigitalOcean Gradient AI running the Qwen3-32B model — we chose Qwen specifically for its strong Indonesian language understanding, which outperformed other models on our legal QA benchmarks
- Architecture: Streaming chat API with a structured intake form for case details, an evidence tracker component, a takedown template generator with platform-specific field mapping, and a report generator that extracts and cites relevant law articles
- Deployment: DigitalOcean App Platform
Challenges we ran into
Qwen3-32B's thinking mode. The model sends reasoning_content chunks before the actual content response. Our initial streaming handler displayed the internal reasoning to users, which was confusing. We had to build a custom streaming parser that filters reasoning tokens and only surfaces the final response.
Language detection tug-of-war. Because our system prompt is loaded with Indonesian legal context (law names, article numbers, Indonesian legal terms), the model defaulted to responding in Bahasa Indonesia — even when users wrote in English. We solved this with explicit language detection logic in the system prompt and response formatting instructions.
Sourcing reliable Indonesian law. Digital versions of UU Hak Cipta and UU Merek are scattered across government PDFs, university repositories, and legal blogs of varying accuracy. We had to manually verify article text against official gazette versions.
Tool vs. chatbot identity crisis. Early versions felt like "just another legal chatbot." The breakthrough was the takedown template generator — when we mapped real form fields for Tokopedia and Shopee's IP complaint processes, it became a tool that produces something you can actually use, not just information you have to act on yourself.
Accomplishments that we're proud of
- Pre-filled takedown templates for 5 major platforms with real form field mapping — not generic letters, but outputs matched to each platform's actual complaint submission process
- 97.6% accuracy on legal QA testing — we evaluated across 5 real-world scenarios with 25 assessment criteria covering legal accuracy, practical advice quality, and citation correctness
- Evidence tracker that persists across sessions — creators can return to their case and pick up where they left off
- Honest enforcement reality checks — we don't oversell what Indonesian IP law can deliver. The system acknowledges that fewer than 10 IP cases per year reach Indonesian courts and steers users toward practical remedies (platform takedowns, DJKI complaints) that actually get results
What we learned
Indonesian IP law is better than its reputation — on paper. UU Hak Cipta 28/2014 is actually comprehensive and grants automatic protection. The problem is enforcement: court cases are rare, expensive, and slow.
Practical remedies beat legal theory. The most impactful thing we can do for a creator isn't explaining Article 113 penalties — it's generating a Tokopedia takedown request they can submit in 10 minutes. We reoriented the entire product around actionable outputs.
A system prompt isn't a moat. Anyone can paste Indonesian law into a prompt. The real value comes from structured workflows — the intake form, evidence checklist, platform-specific templates, and formatted reports. These are the things that turn an AI conversation into a usable tool.
What's next for MineMap
- RAG knowledge base with full-text Indonesian law documents (UU Hak Cipta, UU Merek, PP No. 56/2021) for precise article-level retrieval instead of relying on model knowledge
- Evidence file upload with EXIF metadata extraction to automatically establish creation dates as proof of original authorship
- Case persistence with timeline reminders — follow up with creators on takedown response deadlines and escalation steps
- Southeast Asian expansion — adapt the platform for Thailand, Vietnam, and the Philippines, serving a combined 700M+ population underserved by affordable IP tools
- DJKI database integration for real-time trademark search, so creators can check registration status before filing complaints
Built With
- daisyui
- digitalocean
- gradient-ai
- nextjs
- tailwindcss
- typescript

Log in or sign up for Devpost to join the conversation.