📑 Patents to Markdown for RAG

Convert patents (US/EP/WO) into clean, chunked Markdown for RAG and LLM pipelines via Google Patents — abstract, claims, description.

🌐 The NexGenData Global IP & Trademark Suite

One office is never enough for IP due diligence — search trademarks and patents across the US, EU, China, Japan, Korea, Hong Kong, and WIPO. Every actor below is part of the NexGenData Global IP & Trademark Suite, so a clearance, freedom-to-operate, or brand-protection workflow that starts in one office can extend to every office, worldwide.

Trademarks

• USPTO Trademark Search (US) — word-mark, owner & status lookup across the full USPTO TESS dataset.
• EUIPO Trademark Search (EU) — EUIPO + TMview network across 35+ EU/EEA registries.
• Hong Kong Trademark Search (APAC) — the Hong Kong IPD register by mark, class, owner or status.
• Korea KIPO / KIPRIS Plus (KR) — Korean patents, trademarks & designs via the official KIPRIS Plus API.
• Japan JPO / J-PlatPat (JP) — Japanese trademarks, patents, utility models & designs from J-PlatPat.

Patents

• CNIPA China Patent Search (CN) — Chinese patents from the CNIPA database for IP & innovation tracking.
• Japan JPO / J-PlatPat (JP) — Japanese patents, utility models & designs from J-PlatPat.
• Korea KIPO / KIPRIS Plus (KR) — Korean patents & utility models via the official KIPRIS Plus API.
• WIPO PATENTSCOPE (Global) — worldwide PCT & national patents across WIPO's global collection.
• USPTO Patent Search (US) — US patents with full claims text for prior-art & freedom-to-operate work.
• Patents → Markdown for RAG (AI) — clean Markdown export of patents, ready to embed in a RAG / LLM pipeline. ← you are here

Search every IP office, worldwide — one suite, pay-per-result, structured JSON.

⚡ What you get

One row per chunk: source, url, title, chunkIndex, totalChunks, markdown (LLM-ready, source URL = citation).

🎯 Use cases

1. RAG over this content 2. Vector-store ingestion 3. Searchable knowledge bases 4. Citation-tagged LLM data

🚀 Sample inputs

{ "items": ["US10000000B2", "US9876543B2"], "chunkWords": 800 }

📦 Sample output

{ "source": "US10000000B2", "title": "...", "chunkIndex": 0, "totalChunks": 8, "markdown": "# ...\n..." }

📊 Sample Output

🛠 How it works

1. Fetch each source. 2. Isolate the main document. 3. HTML → ATX Markdown. 4. Chunk ~chunkWords. 5. One row/chunk + citation.

🔗 Related Actors

• USPTO Patent Search\n- Trademark Search\n- SEC Filings to RAG\n- RAG Web Browser

💰 Pricing Example

Pay-per-event: $0.005 per run + $0.04 per document/chunk (document-record).

Chunks	Cost
100	~$4.00
500	~$20.00
2,000	~$80.00
Apify's $5 free credit covers ~124 chunks. Start free →

⚖️ Legal & data sources

Fetches publicly-accessible documents with an identified User-Agent; output includes source URLs for attribution.

❓ FAQ

Citations? Yes. Chunk size? chunkWords. Fresh? Live. Key? No. Inputs? Public HTML. Dedup? Per run.

🆘 Troubleshooting

• Empty markdown → JS-rendered/restricted page. - Boilerplate → use the canonical URL. - Huge → lower inputs/chunkWords. - 404 → check the URL/ID.

🏷️ About NexGenData

Public-data tools for analysts, developers, and operators. thenextgennexus.com