Pascal Pixel

15.3K posts

Pascal Pixel banner
Pascal Pixel

Pascal Pixel

@PascalPixel

👨🏻‍🏭 Independent Designer and Coding Person 😺 Product Hunt’s Ultimate Maker of the Year 🐴 Creator of Horse, the Browser with Trails; https://t.co/QGe8A2FBtw

Lisbon, Portugal Katılım Eylül 2008
507 Takip Edilen12.1K Takipçiler
Sabitlenmiş Tweet
Pascal Pixel
Pascal Pixel@PascalPixel·
i’m honoured, though certainly not surprised, that @producthunt awarded me, jack black, as maker of the year
Pascal Pixel tweet media
English
29
0
139
8.8K
MEK.txt
MEK.txt@michaelmicasso·
reject layout, return to 1.0 tetris 🐒
English
5
3
62
1.1K
Ekaeo
Ekaeo@Ekaeoq·
I don’t think I want my building sculptured by a robot. When you get off the metro in Cologne and see the cathedral that took over 600 years to build, you don’t think about how robots could’ve done it faster and cheaper. You’re reminded of the indomitable and relentless human spirit. Yesterday I learned that there are many companies trying to solve cheap sculpturing, and while it is absolutely impressive, I just don’t care about it. I can’t imagine caring about a building that was “cnc-ed”. I will still very much like the end result, and I prefer a future where robots carve marble compared to a future where everything is just glass and concrete, but I don’t think that’s the point of craftsmanship.
Ekaeo tweet mediaEkaeo tweet media
Ekaeo@Ekaeoq

We will absolutely never build like this again. Developers can't even justify the cost of slightly better materials, or genuinely sustainable layout designs to make a building more sufficient and for it to remain more functional over a longer period of time (this usually means less units per floor). What makes you think that, approving millions upon millions of dollars for for skilled artisans to create sculptures that used to take literal centuries to finish? Just imagine walking into a board meeting and saying you want to spend bazillion cash extra on sculptors, stone carvers and craftsmen so that the facade is "oh so beautiful" It will never happen again.

English
10
0
32
2.7K
Pascal Pixel
Pascal Pixel@PascalPixel·
@yacineMTB But if you’re a smelly insufferable know-it-all with no people skills what else are you supposed to be good for?!
English
0
0
1
47
kache
kache@yacineMTB·
Maybe programming is a low intelligence task and a total god awful waste of human time
English
236
132
3.4K
137.5K
Pascal Pixel
Pascal Pixel@PascalPixel·
We hit 100% byte closure on Golden Sun decompilation!
Pascal Pixel tweet media
English
0
0
5
371
Peter Yang
Peter Yang@petergyang·
I think AI psychosis (trying to make sure your agents are working all the time) is probably a worse addiction than social media.
English
130
29
573
38.4K
Pascal Pixel
Pascal Pixel@PascalPixel·
@TomDavenport Probably my favorite it was an entire open world and 80hr of gameplay in a handheld
English
0
0
0
9
Tom Davenport
Tom Davenport@TomDavenport·
@PascalPixel The game it advance game?! My first emulation experience, this was a FANTASTIC rpg
English
2
0
1
27
Ekaeo
Ekaeo@Ekaeoq·
We will absolutely never build like this again. Developers can't even justify the cost of slightly better materials, or genuinely sustainable layout designs to make a building more sufficient and for it to remain more functional over a longer period of time (this usually means less units per floor). What makes you think that, approving millions upon millions of dollars for for skilled artisans to create sculptures that used to take literal centuries to finish? Just imagine walking into a board meeting and saying you want to spend bazillion cash extra on sculptors, stone carvers and craftsmen so that the facade is "oh so beautiful" It will never happen again.
Alexis Ohanian 🗽@alexisohanian

We’ll build like this again

English
82
6
398
64.8K
Dillon Mulroy
Dillon Mulroy@dillon_mulroy·
actually an insane thing for openai’s head of strategy to publicly say
Dillon Mulroy tweet media
Dean W. Ball@deanwball

Some observations on Kimi: 1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run. 2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China. 3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex. 4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business. 5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this. 6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

English
333
890
11.3K
1.3M
Rinny
Rinny@GamerRinoa·
What's everybody's favorite GF.... Me personally I think I like Shiva's design the best
Rinny tweet media
English
80
39
581
23K
Pascal Pixel
Pascal Pixel@PascalPixel·
It had extracted the djinn battle sprites so I stitched them together for the README.md 📄
GIF
English
0
0
8
621
Pascal Pixel
Pascal Pixel@PascalPixel·
@ds_melon "To meet my expectations, it'd have to shed all the AI code." welp, there go the tokens!
English
0
0
0
165
Pascal Pixel
Pascal Pixel@PascalPixel·
@alexgroberman I can’t rank browser.horse for the life of me, you told me once I was a few good pages away from traffic and hundreds of pages and attempts later… I can’t do it!
English
1
0
0
176
Alex Groberman
Alex Groberman@alexgroberman·
Now that they're adding Fable 5 to Claude's normal subscriptions, it's worth remembering something important about it: Buried inside thousands of lines is one of the clearest explanations we've seen of how Claude decides which websites to search, open, cite and recommend. It also explains why some businesses repeatedly appear in Claude while others don't. Let’s go through it. To see whether Claude, ChatGPT, Google AI, Perplexity and Grok are recommending your business right now, start here. It’s free: seo-stuff.com/free-audit So just for clarity, Anthropic did not publish this as an official Claude ranking guide. The prompt was extracted and published by a public system-prompt archive, so while I would not treat every instruction as a permanent ranking factor, the search process is specific enough to reveal how Claude discovers and evaluates sources. The first major revelation is fairly straightforward: when Claude searches the web, its search tool returns 10 highly ranked results. That is the initial pool Claude receives for that individual search. The prompt does not say these are specifically Google’s top 10 results, and Claude can run multiple searches using different queries. But it does confirm that ranked search visibility plays a major role in determining which websites Claude encounters. Claude essentially starts with a small group of pages that a search engine has already ranked highly. So if your page never enters those groups, Claude may never open, evaluate or cite it for that question. This is why traditional search visibility still matters so much for AI search. seo-stuff.com Claude then applies another layer. The prompt tells it to begin with short, broad searches, generally between one and six words, before narrowing the query when necessary. That could mean searches like: Best payroll software Dental marketing agency Business insurance companies CRM for contractors Claude may take one detailed customer question and investigate it through several shorter searches. Your website therefore needs to make it extremely obvious: What your company does Which category it belongs to Who it serves What problems it solves Which products or services it should be compared against You need pages that map to the different ways buyers describe your category, use case, problem and desired outcome. The prompt also instructs Claude to search the live web when a question involves current information, newer products, specific versions, unfamiliar companies or recommendations that may have changed. That means businesses cannot depend entirely on what Claude learned during training. Your current pricing, product information, case studies, comparisons, reviews and third-party mentions all matter. Freshness is especially important for: Prices Product releases Company information Regulations Rankings Current recommendations Basically, a company missing from Claude’s existing knowledge can still become one of the sources it discovers today. This is where SEO Stuff’s done-for-you package becomes relevant: seo-stuff.com/gold-plan-pack… It combines 10 AI search optimized pieces of content with three DR50+ contextual PR backlinks. The content gives Claude more category, use-case and comparison pages to discover. The backlinks help those pages compete inside the ranked search layer Claude uses to build its source pool. But ranking is only the first step. Claude’s prompt says search snippets are often insufficient. After identifying a potentially useful result, Claude is instructed to open the page and retrieve its full content. The page still has to provide usable evidence. That means: Question-based headings Direct answers Specific definitions Comparison tables Dates Named products Clear authorship Original research Transparent sourcing Pages filled with vague marketing language give Claude very little to work with, whereas pages organized around specific questions and factual answers give it passages that can actually support a response. The system prompt also tells Claude to favor original sources. Examples include: Company websites Official documentation Government sources Peer-reviewed research Primary reports That means your own website can absolutely become the source Claude cites, but it needs to contain original information worth citing. That could include: Actual pricing Product specifications Original research Industry data Customer results Detailed methodologies Company policies Case studies Expert explanations Documentation A generic article repeating information already published across dozens of competing sites gives Claude very little reason to choose your page. Originality becomes even more valuable when Claude is researching an open-ended recommendation. A simple factual question may require one search. A recommendation or comparison can trigger several searches across different categories, products and sources. A connected group of relevant pages gives Claude several ways to encounter your business during the same research process. Claude uses ranked search results for discovery, but it does not automatically trust every commercial page it finds. Also, this probably goes without saying, but a page claiming that your company is the best option is not enough. Your website can explain what you offer, who you serve and why the product is different. Trusted third-party sources help verify those claims. That is why external authority still matters. Relevant press coverage, industry mentions, expert quotes, reviews, podcasts and backlinks can all create additional evidence around the brand. The practical goal is to become visible inside the ranked systems Claude searches while building enough third-party corroboration to make your claims credible. The prompt also says Claude must cite specific claims drawn from web searches. This has major implications for how content should be written. Claude needs passages that answer individual questions, such as: When was the product launched? How much does it cost? Who is it designed for? What feature makes it different? What measurable result did a customer achieve? What evidence supports the company’s claim? The easier those answers are to locate, the easier it becomes for Claude to cite the page. I think of these as claim-sized content blocks. The structure is simple: One clear question. One direct answer. One supporting fact. One source or piece of evidence. That is far more useful to Claude than 1,500 words of general brand copy. Also, Claude can personalize recommendations using information it knows about the user, including their preferences and interests. The company it recommends may change depending on: Industry Company size Location Budget Use case Technical requirements Previous preferences This is why targeting a broad phrase like “best software” or “best agency” is rarely enough. You need pages that clearly explain who the offering is best for. For example: Best accounting software for agencies Best insurance provider for multi-location businesses Best CRM for home service companies The more clearly your pages connect the offering to a specific buyer, the easier it becomes for Claude to justify recommending it to that person. All of this is the system SEO Stuff was built around: seo-stuff.com And to see whether your business is already being recommended across Claude, ChatGPT, Google AI, Perplexity and Grok, check here: seo-stuff.com/free-audit
Alex Groberman tweet mediaAlex Groberman tweet mediaAlex Groberman tweet mediaAlex Groberman tweet media
Alex Groberman@alexgroberman

A company just used AI Search to increase its qualified leads by 1,850%. Those leads also converted at up to 3x the rate of traditional Search. For SaaS companies, agencies, professional services firms, financial companies and pretty much any B2B business trying to increase traffic and sales, this is a really important case study. Let’s go through it. By the way, you can see whether your business is appearing across Google AI, ChatGPT, Claude, Perplexity and Grok here. It’s free: seo-stuff.com/free-audit HubSpot recently broke down the strategy it used to become the most visible CRM across AI platforms. According to the company, qualified leads from AI-generated answers increased 1,850% between Q1 2025 and Q1 2026. Oh, and AI citations increased 433%. Also, those leads converted at up to 3x the rate of traffic from traditional Search. The strategy is worth studying because it lines up very closely with what we have been seeing among SEO Stuff (seo-stuff.com) customers across Google AI, ChatGPT, Claude, Perplexity and broader AI Search. The main takeaway is pretty simple: HubSpot built content and authority around the questions customers ask while deciding which software to buy. Awareness questions. Industry questions. Comparison questions. Pricing questions. Feature questions. Integration questions. Review questions. And questions about whether HubSpot was right for a specific type of business. That last part became particularly important, because HubSpot found that AI systems knew the company, but often struggled to find pages explaining whether the product was right for a specific industry or customer. So HubSpot started creating much more specific industry pages. This is akin to what SEO does with this package: seo-stuff.com/premium-conten… Instead of only explaining what its CRM does, the company created pages explaining how the product works for construction companies, manufacturers, retailers and other individual categories. The pages were built using HubSpot’s existing customer stories, product information and case studies. They included clear answers, structured information and frequently asked questions. 92% of those industry pages were eventually cited by AI answer engines and HubSpot says they produced a 49% increase in AI visibility. That is a pretty strong argument for creating content around specific customers instead of publishing another generic article about the category. A construction company does not only want to know what a CRM is, it wants to know whether the CRM can handle construction sales cycles, multiple projects, field teams, contractors and long follow-up periods. A financial services firm wants to know whether the platform supports its workflows, reporting requirements and compliance needs. An agency wants to know whether it can manage leads, clients, campaigns and reporting without paying for a complicated enterprise system. Those are the questions that move someone closer to buying, and they also give AI systems much more useful information when deciding which company to recommend. HubSpot created comparison content for specific industries too. Those pages reportedly generated a 642% increase in citations and a 58% increase in overall brand mentions. Again, the specificity matters. “Best CRM software” is incredibly broad. “Best CRM for a construction company with a small sales team” gives the AI system a clear customer, use case and decision. The company also updated existing product pages. Headlines were rewritten around real customer questions. FAQs were added. Dense explanations were reorganized using tables and lists. Product information was connected more clearly to the problems customers were trying to solve. HubSpot says those changes increased citations to its product pages by 56%. None of this required abandoning SEO, mind you, as the company still needed useful pages that could be crawled, indexed, trusted and discovered through Google. A lot of companies are treating AI Search like a completely separate marketing channel, but HubSpot connected the entire process. It tracked the commercial questions customers were asking. It identified where competitors appeared instead. It created pages that filled the missing information. It improved existing product content. It built visibility across third-party publishers already being cited by AI. Then it measured citations, visibility, leads and conversion rates. The third-party authority piece was especially important. HubSpot looked for publishers that were already being cited for relevant customer questions but were not mentioning HubSpot. It then worked to increase its presence across those sources. That gave ChatGPT, Perplexity and other AI systems information about HubSpot from websites outside HubSpot’s control. Which makes sense, because every business says its own product is great. AI systems need outside confirmation before confidently recommending one company over another. Relevant articles. Industry publications. Comparison pages. Reviews. Customer discussions. Expert mentions. Credible backlinks. The stronger and more consistent that public footprint becomes, the easier it is for AI systems to understand where a company belongs. This is where HubSpot’s strategy maps directly to smaller businesses. A SaaS company can publish pages for its most valuable industries, use cases, integrations and customer types. An agency can clearly document the industries it serves, the problems it solves, the process it follows and the results it has produced. A financial services company can publish detailed pages covering qualifications, methodology, fees, risks and specific customer situations. A local business can create stronger service and location pages supported by reviews, local citations and third-party authority. The exact pages change by business, but the principle stays pretty consistent. Give Google and AI systems clear information about who you help, what you offer and why a customer should choose you. Then support those claims with credible information across the web. HubSpot’s results also make an important point about traffic quality. AI Search may not always send as many clicks as a traditional Google ranking, but the people who do click may arrive much further into the buying process. They have already asked about the category. They may have compared several companies. They may have narrowed the options by industry, budget or feature. They arrive on the website looking for confirmation. So the visitor may be lower volume, but they can also be much more qualified. This is where SEO Stuff (seo-stuff.com) can help. The done-for-you package combines 10 AI-search-optimized articles with three DR50+ authority placements: seo-stuff.com/gold-plan-pack… The content helps your business cover the questions customers ask before buying, including industries, use cases, comparisons, alternatives, pricing, objections, product details and implementation. The authority placements help confirm what your company says about itself across trusted websites that Google and AI systems use to understand categories. And again, if you're curious about whether your business is appearing across Google AI, ChatGPT, Claude, Perplexity and Grok, you can check here. It’s free: seo-stuff.com/free-audit

English
5
43
124
34.1K
Peer Richelsen
Peer Richelsen@peer_rich·
you can skip a full year of AI and be fine skills harness loops graphs whatever AI skills you acquire will be outdated in 12 months
◢ Brøck Whitten@sintaxi

English
14
5
387
27.4K
Obsolete Sony
Obsolete Sony@ObsoleteSony·
The beauty we threw away in the name of progress
Obsolete Sony tweet media
English
28
164
1.6K
32.4K
MEK.txt
MEK.txt@michaelmicasso·
@PascalPixel Made me want to replay Golden Sun LA again
English
1
0
4
249
Fons
Fons@fonsvandamme·
@PascalPixel We can talk about that after you filed my taxes.
English
1
0
0
24
Pascal Pixel
Pascal Pixel@PascalPixel·
I have had the craziest 6 weeks of productivity in my life. AI did everything. Including my taxes. I am so going to jail. 🥺
English
3
0
14
1.5K