Top all-time · #16223
Pulse AI
Company blog · runpulse.com/blogaidata
Pulse AI is the company blog at runpulse.com, writing about ai and data. Since 2025, 4 different Hacker News readers have linked it 18 times, most often “Why LLMs still have problems with OCR”. No feed was found, so read it on the site itself.
Too few sampled comments (3) to say how often HN recommends it.?
Why HN links it
Linked in discussions of OCR, with posts arguing that LLM-based OCR is unreliable and hallucinates. Commenters cite them as background rather than evaluating the writing.
Summarised by Claude from a sample of 3 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Transform complex PDFs and scans into verified structured data. Enterprise-grade document processing with SOC 2, HIPAA, and GDPR compliance.
Most-linked posts?
- Why LLMs still have problems with OCR3 people · 6 mentions · first linked 2025-02-06
What Hacker News says?
as builders in this space, we decided to put it to the test on complex nested tables, pie charts, etc. to see if the same VLM hallucination issues persist, and to what degree. while results were promising, we found several critical failure nodes across two document domains. check out our blog post here! https://www.runpulse.com/blog/beyond-the-hype-real-world-tes...
Comment linking /blog/beyond-the-hype-real-world-tests-of-mistrals-ocr
on “Mistral OCR”Submitted Why LLMs still have problems with OCR as a story · 218 points
HN discussionritvik here from pulse. everyone’s pretty much made the right points here, but wanted to emphasize that due to the llm architecture, they predict “the most probable text string” that corresponds to the embedding, not necessarily the exact text. this non-deterministicness is awful for customers deploying in production and a lot of our customers complained about this to us initially. the best approach is to build a so…
Comment linking /
on “Ingesting PDFs and why Gemini 2.0 changes everything”Submitted Putting Andrew Ng's OCR models to the test as a story · 124 points
HN discussionIt's definitely a possibility if there's a point where LLM-based OCR is applied. See https://www.runpulse.com/blog/why-llms-suck-at-ocr and its related HN discussion https://news.ycombinator.com/item?id=42966958
Comment linking /blog/why-llms-suck-at-ocr
on “iPhone 16 cameras vs. traditional digital cameras”Submitted Beyond the Hype: Real-World Tests of Mistral's OCR as a story · 4 points
HN discussionhey -- wrote a blog post about this exact phenomena [1] (also posted on HN couple weeks back [2]). tldr: maintaining confidence from LLM nondeterministic outputs over millions of pages is a problem. especially in production environments like healthcare, finance, etc. we've noticed decently high hallucination rates, even in finetuned LLMs. [1]: https://www.runpulse.com/blog/why-llms-suck-at-ocr [2]: https://news.ycom…
Comment linking /blog/why-llms-suck-at-ocr
on “Replace OCR with Vision Language Models”Submitted PulseBench-Tab: Open-source, multilingual benchmark for table extraction as a story · 3 points
HN discussionour current customers are both enterprises and individuals. pricing page is here https://www.runpulse.com/pricing-studio-pulse
Comment linking /pricing-studio-pulse
on “Why LLMs still have problems with OCR”Submitted Computational Complexity of Schema-Guided Document Extraction as a story · 2 points
HN discussionhey, i recommend checking out the previous HN threads [1] on why LLMs shouldn’t be used in production-grade OCR, especially if accuracy is super important (as in the audiobook case) we wrote the post and created Pulse [2] for these exact use cases, feel free to reach out for more info! [1]: https://news.ycombinator.com/item?id=42966958 [2]: https://runpulse.com
Comment linking /
on “Ask HN: What is the best method for turning a scanned book as a PDF into text?”Submitted Why Tables Are the Hardest Problem in Document AI as a story · 2 points
HN discussionclaude is definitely better than gpt -- but both have their flaws! they pretty much fall flat on their face with nested entries, low-fidelity images, etc. (we detailed this heavily in our blog post here [1]) other ocr providers are doing a great job - we personally believe we have the highest accuracy tool on the market. we're not here to dunk on anyone just provide unbiased feedback when putting new document extrac…
Comment linking /blog/why-llms-suck-at-ocr
on “Putting Andrew Ng's OCR models to the test”Submitted Evaluating Document Extraction Accuracy as a story · 2 points
HN discussionwe disagree! we've found llms by themselves aren't enough and suffer from pretty big failure modes like hallucination and inferring text rather than pure transcription. we wrote a blog about this [1]. the right approach so far seems to be a hybrid workflow that uses very specific parts of the language model architecture. [1] https://www.runpulse.com/blog/why-llms-suck-at-ocr
Comment linking /blog/why-llms-suck-at-ocr
on “Launch HN: Pulse (YC S24) – Production-grade unstructured document extraction”Submitted Why Semantic Understanding Breaks at Page Boundaries as a story · 2 points
HN discussionAre you aware that AI-based OCR very often hallucinates? https://www.runpulse.com/blog/why-llms-suck-at-ocr
Comment linking /blog/why-llms-suck-at-ocr
on “Ask HN: Who's regularly using LLMs at work?”Submitted Legacy OCR Tools Are Failing the Legal Industry: Here's Why as a story · 2 points
HN discussion