chelm

168 meters above sea level

I can make this more transparent; it's the same issue that Parashift had, which ran https://intelligentdocumentprocessing.com/, which they terminated a month ago.

IDP is not a really sexy market. There are only a few people, who are working in the industry.

I do this in my free time to give small vendors a chance, as big corporates like Rossum, Abbyy, or Kofax (now Tungsten) just rule the market by their ads spent.

I can also make it closed source and ask for a fee to get listed as Gartner would do it in their IDP Magic Quadrant.

I did spend 1/580 of the time on the page konfuzio. Ok, true. And I spent 579/580 on the market. https://idp-software.com/sitemap.xml

tl;dr: years ago, Tesseract was the go to tool to extract text. Nowadays, vLLMs can not only extract the text and the layout but also context and provide structured data or even interpret or map data across documents. Prices dropped significantly, while extraction, classification and modification capabilities increased.

The intelligent document processing (a funny marketing term on top of OCR) market moves from "Can software extract the text", which is normally measured by benchmarks, to can software autonomously run "a" specific company process.

the fallback is called human in the loop, hallucination (LSTM vs. vLLM), prompt engineering.

proof me wrong: the hardest challenge is no longer the OCR accuracy but the integration and issue handling in production. Probably "an agentic team can handle this" ^^

ahahah, probably not. Looking at my own handwriting: Neither in writing nor in reading.

I find it interesting how the prompt changes the result.

If you let the model focus on the text, the open source got so good in the last year. That's remarkable. When you change to prompt to not only extract the text but also extract specific information, the pure text extraction result gets worse. For me, it worked to run two prompts on the same document to get both in a meaningful accruacy.

I linked your board already. You are right.

Do you know a benchmark that tries to measure the bussines accuracy.

Most benchmarks focus on the charackter level.

IDP Software typically uses metadata to map information that is either not readable or missing in the document, e.g. extracting the VAT and mapping the street, house number, cip and city.

I think there are many models and many providers. However, it's really difficult to measure the accuracy on a porcess not just on a character level.

https://idp-software.com/vendors/nanonets/

I saw that the leaderboard is hosted by Nanonets. Totally fine for me. So you might be the expert about Nanonets: Let me know if you want to update your post on my site.

https://www.zeit.de/digital/datenschutz/2026-04/vibe-coding-...

Funny to see this approach trending! I published this a month ago.

https://wire.wise-relations.com/guides/components/

my takeaway:

- add lint or errors, otherwise your formatting will break, e.g. LLMs and humans will add text too long or too short and your design system will not be able to handle this.

- it's great for low token input

- validate the layout of the user vs. the components used.

- seen here before: https://myst-parser.readthedocs.io/en/latest/syntax/optional...

You scrape your screen continuously and OCR it? Never heard of this use case.

I mean "a" text! I was just curious how you write. Do you prefer to write comments?

Ok, let's not discuss the content but the format.

Who has ever had multiple sentences?

Many? https://forum.wordreference.com/threads/two-sentences-in-a-t...

Sources for claims that call for evidence

Absolutely. You got the joke, or? This was the main point of the full article. No primary sources. Only unverified aggregates. Strong contrast to what I did normally once per month.

Variable paragraph lengths

I tried to compare it to the URL you posted. It's quite similar. I would have rather have said. Shorter sentences. Shorter Paragraphs. But let's not fight on this ;)

IMHO LLMs cannot provide statistically confident measures, and they are terrible at pretending to be capable of doing so.

What worked: You use an OCR that provides character/word-level bounding boxes and let the LLM extract from data. Then the LLM is capable of "calculating" a confidence of extracted data.

I think I made it obvious what the article is about: no boasting, not "copying someone else's homework". Which text did you last publish? Can you be more specific? I would be genuinely interested in specific changes you would do if you were the editor.

Speaking about "that the state of the art tools", might be 6 months or 20 years old. Surfaced opinions might rely on software that a company licensed 2 years ago. Sadly, we need to take this enterprise speed of adaptation into account.

I found only a few that correct OCR by using LLM. I think it feels too risky.

Think of an LLM that corrects 898,00 to 888,00. It feels like the David Kriesel Xerox case. Still, it's an interesting way to think of the issue of optical character recognition.

The relevant term is "bounding box", as you probably need the confidence level of a character or word, not just the image. I built such an interface. I think the effort is only worth it if you really have multi-millions of pages.

Niels lately posted a lot about other OCR engines: https://www.linkedin.com/posts/niels-rogge-a3b7a3127_lots-of...

Pretty balanced take. I think if a human gains information or saves time, it's still worthwhile. Surely, I don't publish those clickbaits. That's AI slop.

Did you read the article?

RAG provided me no way to read the content myself. I now integrate the knowledge into a static page that I can read and edit myself in Markdown. Similar to MkDocs. But after I edit the content or remove elements that are no longer true, I build a JSON file and tell the agent how to query this source.

python -c " import json, wire, pathlib d = json.loads((pathlib.Path(wire.__file__).parent / 'assets/search_index.json').read_text()) [print(e['title'], e['url']) for e in d if 'QUERY' in (e.get('body','') + e.get('title','')).lower()] "

python -c " import json, wire, pathlib d = json.loads((pathlib.Path(wire.__file__).parent / 'assets/search_index.json').read_text()) [print(e['body']) for e in d if e.get('url','') == 'PATH'] "

https://wire.wise-relations.com/use-cases/replace-rag/

I did not write a full function in the last 4 months. But 3 lines the AI just did not get right autonomously.

https://wire.wise-relations.com/news/2026-03-16-three-lines/

I can say I am pleased to see that!

I did something similar with it. It addresses the pain points of typical real estate owners (landlords and tenants). https://der-walter.de/

On Google, the game I made is more trending than the tools.

I think the monetization is still unresolved.

I did some experiments with the File System Access API. So users can select a folder, work in it via the web app, and just leave.

I just made it for my blog and added a few templates in https://github.com/atraining/featured-image-generator-wordpr.... Perhaps someone wants to create images faster and directly add relevant alt text and correctly named media files to the WordPress media library, too.

I wrote an article and would be interested if you think it's a must have or just another hype topic.

Tools like https://mlco2.github.io/codecarbon/ or https://developers.thegreenwebfoundation.org/co2js/tutorials... might provide a staring point.

Cool work! I did not yet add vendors that generate “synthetic data” or collect documents to my list. If you like, create a PR, I'll merge it. https://idp-software.com/contribution/

Adding 4 hypotheses to the dicussion.

B2B opportunities are inherently easier to identify during early-stage evaluations due to clearer revenue models and problem-solution alignment.

YC’s network amplifies B2B growth by directly connecting startups to other batch companies as potential early adopters or customers. I found it interesting that https://www.ycombinator.com/companies/legora (former Leya) is using Reducto. If I get page 9 in the pitch deck correct: https://www.pitchdeckinspo.com/deck/Reducto_02c1f2af-3fa2-4a...

Early-stage B2B startups require less capital and shorter timelines to demonstrate product-market fit compared to B2C.

YC’s founder archetype—technical, execution-driven, and efficiency-focused - naturally gravitates toward building scalable B2B solutions.

Thanks for pointing out Reducto! I added it to my market overview: https://idp-software.com/vendors/reducto-ai/

TLDR

The IDP market remains a massive and growing space. There will be a new segment of the market for simple cases that do not need domain expertise, validation, and integration. Generic Document AI tools, so-called AI wrapper, provide easy wins for basic input extraction / categorization and splitting tasks.

Operational complexity, on-premises, integrating directly with enterprise infrastructure and domain-specific validation across fields mean different workflows require specialized handling. I think this is why Hyland, Abbyy and others can compete with the market, event the tech stack lagging.