The same day, a post on reddit was about: "We built 3B and 8B models that rival GPT-5 at HTML extraction while costing 40-80x less - fully open source" [1].<p>Not fully equivalent to what is doing Skyvern, but still an interesting approach.<p>[1]
<a href="https://www.reddit.com/r/LocalLLaMA/comments/1o8m0ti/we_built_3b_and_8b_models_that_rival_gpt5_at_html/" rel="nofollow">https://www.reddit.com/r/LocalLLaMA/comments/1o8m0ti/we_buil...</a>
This is exactly the direction I am seeing agent go. They should be able to write their own tools and we are soon launching something about that.<p>That being said...<p>LLMS are amazing for some coding tasks and fail miserably at others. My hypothesis is that there is some sort of practical limit to how many concepts an LLM can hold into account no matter the context window given the current model architectures.<p>For a long time I wanted to find some sort of litmus test to measure this and I think I found one that is an easy to understand programming problem, can be done in a single file, yet complex enough. I have not found a single LLM to be able to build a solution without careful guidance.<p>I wrote more about this here if you are interested: <a href="https://chatbotkit.com/reflections/where-ai-coding-agents-go-to-die" rel="nofollow">https://chatbotkit.com/reflections/where-ai-coding-agents-go...</a>
> For a long time I wanted to find some sort of litmus test to measure this and I think I found one that is an easy to understand programming problem, can be done in a single file, yet complex enough. I have not found a single LLM to be able to build a solution without careful guidance.<p>Plan for solving this problem:<p>- Build a comprehensive design system with AI models<p>- Catalogue the components it fails on (like yours)<p>- These components are the perfect test cases for hiring challenges (immune to “cheating” with AI)<p>- The answers to these hiring challenges can be used as training data for models<p>- Newer models can now solve these problems<p>- You can vary this by framework (web component / React / Vue / Svelte / etc.) or by version (React v18 vs React v19, etc.)<p>What you’re doing with this is finding the exact contours of the edge of AI capability, then building a focused training dataset to push past those boundaries. Also a Rosetta Stone for translating between different frameworks.<p>I put a brain dump about the bigger picture this fits into here:<p><a href="https://jim.dabell.name/articles/2025/08/08/autonomous-software/" rel="nofollow">https://jim.dabell.name/articles/2025/08/08/autonomous-softw...</a>
also training data quality. they are <i>horrifyingly</i> bad at concurrent code in general in my experience, and looking at most concurrent code in existence.... yeah I can see why.
The really depressing part about LLMs (and the limitations of ML more generally) is that humans are really bad at formal logic (which is what programming basically is), and instead of continuing the path of making machines that made it harder for us to get it wrong, we instead decided to toss every open piece of code/text in existence into a big machine that then reproduces those patterns non-deterministically and use that to build more programs.<p>One can see the results in a place where most code is terrible (data science is the place I see this most, as it's what I do mostly) but most people don't realise this. I assume this also happens for stuff like frontend, where I don't see the badness because I'm not an expert.
> is that humans are really bad at formal logic (which is what programming basically is),<p>The tricky part is that I don't think all programming is formal logic at all, just a small part. And this thing with that different code is for different purposes really screws up LLMs reasoning process unless you make it really clear what code is for what.
Codex (GPT-5) + Rust (with or without Tokio) seems to work out well for me, asking it to run the program and validate everything as it iterates on a solution. I've used the same workflow with Python programs too and seems to work OK, but not as well as with Rust.<p>Just for curiosities sake, what language have you been trying to use?
Or when code is fully vectorizable they default to using loops even if explicitly told not to yse loops. Code I got a LLM to solve for a fairly straightforward problem took 18 minutes to run.<p>my own solution? 1.56 seconds. I consider myself to be at an intermediate skill level, and while LLMs are useful, they likely wont replace any but the least talented programmers. Even then i'd value human with critial thinking paired with an LLM over an even more competent LLM.
In my experience, because the Clojure concurrency model is just incredibly sane and easy to get right, LLMs have no difficulty with it.
With the upcoming release of Gemini 3.0 Pro, we might see a breakthrough for that particular issue. (Those are the rumors, at least.) I'm sure not fully solved, but possibly greatly improved.
I feel like this is how normal work is. When I have to figure out how to use a new app/api etc, I go through an initial period where I am just clicking around, shouting in the ether etc until I get the hang of it.<p>And then the third or fourth time its automatic. Its weird but sometimes I feel like the best way to make agents work is to metathink about how I myself work.
Yes, it is easy. LLMs have reduced my maintenance work on scraping tasks I manage (lots of specialized high-traffic adfield sites) by 99%<p>What used to be a constant almost daily chore with them breaking all the time at random intervals is now a self-healing system that rarely ever fails.
One of the uses for AI I'm excited about - maintaining systems, keeping up with the moving targets.
Could you elaborate on your setup please?
Interesting. Could you elaborate? Is there a specific reason that it doesn't do 100% of the work already?
That's the dream
We had a similar realization here at Thoughtful and pivoted towards code generation approaches as well.<p>I know the authors of Skyvern are around here sometimes --
How do you think about code generation with vision based approaches to agentic browser use like OpenAI's Operator, Claude Computer Use and Magnitude?<p>From my POV, I think the vision based approaches are superior, but they are less amenable to codegen IMO.
I think they're complementary, and that's the direction we're headed.<p>We can ask the vision based models to output <i>why</i> they are doing what they are doing, and fallback to code-based approaches for subsequent runs
Unrelated, but thoughtful gave us some very very helpful feedback early in our journey. We are big fans!
Off topic, but because the article mentioned improper usage of DOM, I put down the UK government's design system/accessibility. It's well documented, and I hope all governments have the same standard. I guess they paid a huge amount of money to consultants and vendors.<p>[1] <a href="https://design-system.service.gov.uk/components/radios/" rel="nofollow">https://design-system.service.gov.uk/components/radios/</a>
Your example use case is automatically filling out an IRS form, operated by the sort of IRC department that makes a webform that's only up during business hours? Do you realize how legally risky that is to create, and how legally risky that will be to operate?
In AI First workshops. By now I tell them for the last exercise "no scrappers". the learning is to separate reasoning (AI) from data (that you have to bring.) and ai coded scrappers seem a logical, but always fail. scrapping is a scaling issue, not reasoning challenge. also the most interesting websites are not keen for new scrappers.
Not at all in my opinion. Its a zero sum game against anti bot technologies also employing AI to block scrapers.
A point orthogonal to this; consider whether you need browser automation at all.<p>If a website isn't using Cloudflare or a JS-only design, it's generally better to skip playwright. All the major AIs understand beautifulsoup pretty well, and they're likely to write you a faster, less brittle scraper.
The vast majority of the modern internet falls into one of those two buckets though, no?
I mostly scrape government data so the sites are a little 'behind' on that trend, but no. Even JS heavy sites are almost always pulling from a JSON or graphql source under the hood.<p>At scale, dropping the heavier dependencies and network traffic of a browser is meaningful.
IF you can use crawlers, definitely do.<p>They aren't enough for anything that's login-protected, or requires interacting with wizards (eg JS, downloading files, etc)
If.
You gain experience getting interactions with other agencies optimised by dealing with them yourself. If the AI you rely on fails, you are dead in the water. And I'm speaking as a fairly resilient 50 year old with plenty of hands-on experience, but concerned for the next generation. I know generational concern has existed since the invention of writing, and the world hasn't fallen apart, so what do I know? :)
Over the past few days I've spent a lot of time dealing with terribly designed UIs. Some legitimate and desired use cases are impossible because poor logic excludes them.<p>Is AI capable of saying, "This website sucks, and doesn't work - file a complaint with the webmaster?"<p>I once had similar problems with the CIA's World Factbook. I shudder to think what an I would do there.
I'd be all over Skyvern if only they had enterprise compliance agreements available.
this matches our personal experience, too
I tried skyvern like 6 mo ago and it didn’t work for scraping a site that sounds like welp. Ended up doing it myself. Was trying to scrape data across Bay Area.<p>That said I’d try it again but I don’t want to spend money again.
the hardest part of scrapping is bypassing Cloudflare/captchas/fingerprinting etc
I misread this as 'sky scrapers'