Top all-time · #12296
Unstract.com →
Company blog · unstract.com/blog · on HN as naren87aidata
Unstract.com → is the company blog at unstract.com, writing about ai and data. Since 2023, 6 different Hacker News readers have linked it 64 times, most often “PDF Hell: Why Is Extracting Data Still a Nightmare? (2024)”. It is still writing: the last post was on 2026-10-27, and you can follow it by feed.
Too few sampled comments (1) to say how often HN recommends it.?
Why HN links it
The linked pages describe LLMWhisperer, a document OCR tool from Unstract. The one comment questions whether the product's page is accurate about its use of LLMs.
Summarised by Claude from a sample of 1 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Unstract Flow turns a described process into production-grade document automation workflows — deterministic where it matters, agentic where it helps. Highly accurate.
Most-linked posts?
- PDF Hell: Why Is Extracting Data Still a Nightmare? (2024)3 people · 13 mentions · first linked 2024-05-29
- Show HN: LLMWhisperer – Prepare complex documents for LLMs3 people · 12 mentions · first linked 2024-05-09
- Specification Grounding: The Missing Link in Vibe Coding3 people · 5 mentions · first linked 2025-07-10
- Why LLMs Are Not (Yet) the Silver Bullet for Unstructured Data Processing2 people · 8 mentions · first linked 2025-05-06
- Show HN: Unstract – AGPL, LLM-powered Fivetran/Twilio for unstructured docs2 people · 4 mentions · first linked 2023-12-25
- Why deterministic output from LLMs is nearly impossible2 people · 4 mentions · first linked 2025-07-28
- Exploring the Extractive Capabilities of Large Language Models2 people · 2 mentions · first linked 2024-05-10
What Hacker News says?
Why PDF parsing is Hell[1]: Fixed layout and lack of semantic structure in PDFs. Non-linear text flow due to columns, sidebars, or images. Position-based text without contextual or relational markers. Absence of standard structure tags (like in HTML). Scanned or image-based PDFs requiring OCR. Preprocessing needs for scanned PDFs (noise, rotation, skew). Extracting tables from unstructured or visually complex layout…
Comment linking /blog/pdf-hell-and-practical-rag-applications/
on “DeepSeek OCR”Submitted Why deterministic output from LLMs is nearly impossible as a story · 26 points
HN discussionUnstract - https://unstract.com/ They are a month away from launch(both open source and cloud) The team might be able to give you a quick demo on your specific requirements.
Comment linking /
on “Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?”Submitted PDF Hell: Why Is Extracting Data Still a Nightmare? (2024) as a story · 8 points
HN discussionOther Document parsers for use in LLM/RAG that are not mentioned in this thread https://unstract.com/llmwhisperer/ https://unstructured.io/ https://docs.llamaindex.ai/en/stable/llama_cloud/llama_parse... https://www.eyelevel.ai/
Comment linking /llmwhisperer/
on “Show HN: LLM-aided OCR – Correcting Tesseract OCR errors with LLMs”Submitted Specification Grounding: The Missing Link in Vibe Coding as a story · 7 points
HN discussionMost of the document processing automation projects at an enterprise level require parsing of complex documents with tables, forms, handwriting, checkboxes, scanned documents. Example includes ACORD insurance forms, IRS tax forms and bank statements. I’m not even getting into how different each document can be even if they are of the same nature. For any one curious on automating document processing end-to-end by le…
Comment linking /llmwhisperer/
on “Show HN: LLM-aided OCR – Correcting Tesseract OCR errors with LLMs”Submitted PDF Hell and Practical RAG Applications as a story · 6 points
HN discussionOther PDF parsing woes include: 1. Identifying form elements like check boxes and radio buttons. 2. Badly oriented PDF scans 3. Text rendered as bezier curves 4. Images embedded in a PDF 5. Background watermarks 6. Handwritten documents PDF parsing is hell indeed: https://unstract.com/blog/pdf-hell-and-practical-rag-applica...
Comment linking /blog/pdf-hell-and-practical-rag-applications/
on “So you want to parse a PDF?”Submitted Show HN: Unstract – AGPL, LLM-powered Fivetran/Twilio for unstructured docs as a story · 5 points
HN discussionTrue indeed, but there are a few problems — hallucinations and trusting the output(validation). More here https://unstract.com/blog/why-llms-struggle-with-unstructure...
Comment linking /blog/why-llms-struggle-with-unstructured-data/
on “PDF to Text, a challenging problem”Submitted Jev and where a System One model fits in document processing as a story · 4 points
HN discussionPDF parsing is hell indeed, with all sorts of edge cases that breaks business workflows, more on that here https://unstract.com/blog/pdf-hell-and-practical-rag-applica...
Comment linking /blog/pdf-hell-and-practical-rag-applications/
on “PDF to Text, a challenging problem”Submitted PDF Hell and Practical RAG Applications as a story · 4 points
HN discussionI will try it with some complex layout PDFs or documents with tables. These documents have real business use cases for automation — insurance, banking, etc. Anyone here who wants to convert PDF documents or scanned images as it is preserving the layout, do try LLMWhisperer - https://unstract.com/llmwhisperer/
Comment linking /llmwhisperer/
on “MarkItDown: Python tool for converting files and office documents to Markdown”Submitted PDF Hell: Why is extracting data still a nightmare? as a story · 3 points
HN discussionLLMs are not yet there for complex and diverse document parsing use cases, especially at an enterprise scale (processing millions of pages). Some of the reasons are: Complex layouts, nested tables, tables spanning multiple pages, checkboxes, radio-buttons, off-oriented scans, controlling LLM costs, checking hallucinations, Human-in-the-loop integration, and privacy. More on the issues: https://unstract.com/blog/why-…
Comment linking /blog/why-llms-struggle-with-unstructured-data/
on “Don't bother parsing: Just use images for RAG”Submitted Specification Grounding: The Missing Link in Vibe Coding as a story · 3 points
HN discussion> try LLMWhisperer( https://unstract.com/llmwhisperer/ ) for OCR. It avoids LLMs The website you linked says it uses LLMs?
Comment linking /llmwhisperer/
on “Replace OCR with Vision Language Models”Submitted PDF Hell: Why Is Extracting Data Still a Nightmare? (2024) as a story · 3 points
HN discussionThe primary issue with LLMs is hallucination, which can lead to incorrect data and flawed business decisions. For example, Llamaparse( https://docs.llamaindex.ai/en/stable/llama_cloud/llama_parse... ) uses LLMs for PDF text extraction but faces hallucination problems. See this issue for more details: https://github.com/run-llama/llama_parse/issues/420 . For those interested, try LLMWhisperer( https://unstract.com/ll…
Comment linking /llmwhisperer/
on “Replace OCR with Vision Language Models”Submitted Specification Grounding: The Missing Link in Vibe Coding as a story · 3 points
HN discussionThe primary challenge is not just about harnessing AI for search; it's about preparing complex documents of various formats, structures, designs, scans, multi-layout tables, and even poorly captured images for LLM consumption. This is a crucial issue. There is a 20 min read on why parsing PDFs is hell: https://unstract.com/blog/pdf-hell-and-practical-rag-applica... To parse PDFs for RAG applications, you'll need too…
Comment linking /llmwhisperer/
on “Ask HN: I have many PDFs – what is the best local way to leverage AI for search?”Submitted Why LLMs Are Not (Yet) the Silver Bullet for Unstructured Data Processing as a story · 3 points
HN discussionAny one here exploring to solve extraction/parsing problem for RAG, do try LLMWhisperer[1]. Try it with complex layout documents -> https://pg.llmwhisperer.unstract.com/ If anyone wants to solve for RAG right from loading from source, extraction, and sending processed data to destination/API, try Unstract [2] (it is open-source) [1] https://unstract.com/llmwhisperer/ [2] https://github.com/Zipstack/unstract
Comment linking /llmwhisperer/
on “Show HN: R2R V2 – A open source RAG engine with prod features”Submitted Comparing Approaches for Using LLMs for Structured Data Extraction from PDFs as a story · 3 points
HN discussionIf you want to do document OCR/PDF text extraction with decent accuracy without using an LLM, do give LLMWhisperer[1] a try. Try with any PDF document in the playground - https://pg.llmwhisperer.unstract.com/ [1] - https://unstract.com/llmwhisperer/
Comment linking /llmwhisperer/
on “Show HN: Zerox – Document OCR with GPT-mini”Submitted What we learned building a multi-agent PDF table extractor as a story · 2 points
HN discussionAnother common stack that is commonly use is Langchain + Pydantic https://unstract.com/blog/comparing-approaches-for-using-llm...
Comment linking /blog/comparing-approaches-for-using-llms-for-structured-data-extraction-from-pdfs/
on “Sampling and structured outputs in LLMs”Submitted Why LLMs Are Not (Yet) the Silver Bullet for Unstructured Data Processing as a story · 2 points
HN discussion