@dr Worked with him and he’s genuinely a good guy. He’s got a solid team of people he partners with. Don’t underestimate how big the internet is and what numbers are possible 🌏
English
Andrew Hogan
107 posts
Andrew Hogan
@ahogan
Ex-professional turned amateur.
Katılım Ağustos 2007
129 Takip Edilen1.4K Takipçiler
If my products were at +$900K MRR I don’t think I’d be writing a weekly newsletter
English
@tinyprojectsdev This is awesome. Perfect use case for stablecoins. Nice work Ben!
English
First post in 4 years
I’ve been busy bootstrapping PromptBase to 450k+ users
But I was burning $9,400/mo in opaque Stripe fees for seller payouts
So today I’m launching Zoneless: an open-source clone of Stripe Connect using USDC
Identical API, except payouts cost $0.002
I’ve been dogfooding it on PromptBase for 3 months:
- 2,200+ sellers onboarded
- 1,400+ payouts completed
- $9.4k/mo in Stripe fees → ~$5/mo
- 73% of sellers chose Zoneless over Stripe
English
No need for ‘you're the expert’ prompts. Biggest win is making Claude a Paperbell expert vs goo ads expert. Give Claude campaign data (esp. search terms report and ad copy). Twin this with business context via skill or cluade.md file on product features/use cases and customer personas. Ask it to review account structure.
English
How are we doing "you're the expert" type prompting with opus 4.6?
For help with google ads for example - do I need to build a skill? Put in a "you're a google ads expert" type of prompt? Or just ask my question?
English
🥦You can create your own plate, discover easter eggs and unlock new foods at eat-real-food.netlify.app
🍅 Note the the food values are for demo purposes only - they will differ depending on the portions and specific foods you consume.
⚠️ Disclaimer: This is a demo site and not affiliated with the USDA. I'm not a professional nutritionist (and the values in this demo have not been officially approved) but I would be happy to work with an expert to make this an interactive educational experience.
English
Reimagined the brand new realfood.gov website as an interactive drag and drop game 🇺🇸🥦🍞🥩
🥕 Mix and match from the different foods
🍽️ Unlock new foods by creating a "balanced" plate
📸 Share your customizations by downloading and posting a pic (1.3M+ combos)
English
English
@LyalinDotCom @NanoBanana Awesome! I’m actually in the middle of writing a Skill to use Nano Banana, and this fits perfectly with what I’m building.
Thanks… I’ll give it a try and let you know how it goes.
English
MCP has it's use... but often I'm so tired of configuring MCP servers for what is becoming a common use case for my side projects, using @NanoBanana to generate images using my coding agent
So I've created Nano Banana CLI, more portable + agent friendly
github.com/LyalinDotCom/n…
English
You can publish any artifact and share them with team members, clients etc
The can be custom designed and fully branded.
Here a link to my example: claude.ai/public/artifac…
English
Claude Artifacts do not get enough attention
Skills + Artifacts replace any dashboard tool
This example was created in minutes. Claude clustered the top SEO keywords from Wayfair website, created a tree diagram AND added key insights.
English
@levelsio @PascalPixel I remember you posting about the 'modern photo shoot' over a year ago. You did it. Nice one! x.com/levelsio/statu…
@levelsio@levelsio
2 years ago I knew AI would change photography but didn't really know how Now it's becoming much clearer to me where it's going: A modern photo shoot will be anywhere from hybrid to 100% AI depending on the budget It will be combining different trained concepts: people, clothes, places, light, styles, brand style guides (eg Nike) and it'll be able to generate/iterate lightning fast (eg try different product/ad ideas before producing them to validate them) 🙂 A non-existing 100% AI photo model (person) will be used, or if budget allows a real life famous model (like an influencer) is licensed via an agency which provides the dataset or LoRa (but more importantly the legal rights) There'll be a big legal industry fighting companies using influencers likeness (as there kinda is now already) because that'll be easier with AI 👗 For clothes: try on models are getting very close to perfect, I'd say they're 80% there now. That means any model can wear any clothing that hasn't been worn yet (so just a pic of the clothing from a design app or factory is enough for a model to wear it) 🏡 For places: you might see dataset providers pop up to sell pre-made finetuned models of places like Amsterdam because they're not SPECIFICALLY accurate in the current foundational models, or even indoor spaces 💡 Light/styles: see how companies now sell Lightroom presets, I am not sure there'll be a market for it, but you'll need lots of finetunes on analog film and styles (CivitAI of course already does this for free) ⚡️ Generating content will be lightning speed and can be highly automated too, that means companies can try different products/content and see what sticks before producing them It's hard to predict but if you combine these components you might have the new way commercial/advertisement/fashion photography is done in the future Just my ideas in a rant, might be wrong 😀
English
✨ After many years, I finally got multi characters working on Photo AI now 🥳🥳🥳
You can now train and then select up to 4 multiple characters (by holding [ ⇧ Shift ] or long pressing on mobile) and they will be generated all together in a single photo
This makes Photo AI much more interesting as a product now because most photos are not of a single person, but of couples, or groups like families (think how much group family photographers charge etc)
This was the #1 most requested feature for years and was never possible because with both Stable Diffusion and Flux, I'd train a LoRa on one character (person), and if I combined multiple LoRas, instead of generating both people it would merge the two people's likeness into one, so absolutely useless. For a long time I tried to think of other ways, like generating one image first with multiple people, then replacing those people with each LoRa with img2img, but it was all too unwieldy and didn't really work well
Couples kinda worked if you'd train a new model with photos of both of you, but every photo had to kinda have the same position like (you left, partner right) or it'd start messing it up, TL;DR the models were too dumb to do this
But with all the new models now like Nano Banana Pro and Hyper Realism, they're finally smart enough to understand what you want to generate
The results are very detailed and you can do stuff like
1) extreme close ups with cool lights (AI Photography pack)
2) put you and your gf in Berghain (Berlin photo pack)
3) turn you guys into digital nomads in Bangkok (digital nomad pack)
THANK YOU FOR YOUR ATTENTION TO THIS MATTER 😊
@levelsio@levelsio
🍌 Nano Banana Pro can do GTA: San Andreas pics very well And with Photo AI you can put yourself in it 🤠
English
@trq212 Spent more time with claude code than my family over the holidays. It's unreal. Thanks for sharing these.
English
For anyone interested this skill connects to the pytrends python package. Excited to create more skills. I'm finding them incredibly useful.
English
Google Trends meet Claude!
Made this 'skill' so I can chat to Google trends.
Works better than expected. Love the actionable insights!
English
@earthlingworks Agreed. Outsourcing the distribution bit is quickly becoming a non option.
English
I know a lot of people probably feel the same.
Maybe it’s a way of framing things, but to me (in my head) building isn't complete without the marketing/distribution.
I’d treat it that way and not move on till it’s done (by me or someone else).
Josh Pigford@Shpigford
Real talk: I’m struggling with how to market all the stuff I’m building. I know it needs to be done and it’s easy to ask ChatGPT for ideas, but ultimately I kinda feel paralyzed.
English
@Shpigford You know what to do. Business mode > tinkering mode. baremetrics.com/blog/marketing…
English
Real talk: I’m struggling with how to market all the stuff I’m building.
I know it needs to be done and it’s easy to ask ChatGPT for ideas, but ultimately I kinda feel paralyzed.
English
@RicWomersley @andrewglynch Me too, really enjoyed income articles. 2025 comeback ?
English
@andrewglynch Really miss your net income articles.
Learnt soo much as a total accounting novice.
Hope you do more when family life is a bit less intense 🍼
English
Last summer, I wrote an article breaking down the business model and valuation of Center Parcs, a leisure and hospitality business here in the UK.
Last week, that article hit the front page of Hacker News. So obviously I read all the comments.
Some of them were wonderful! Very complimentary. Very mindful. Very demure.
And some of the comments were RIDICULOUS.
(I'll link the comments and the article below).
Even among a community of people as smart as Hacker News readers, there's such a misunderstanding of how different business models impact a company's P&L, balance sheet and cashflow.
E.g. comments like:
"Center Parcs is overpriced, which is why their EBITDA margins are so high."
Wrong. Their EBITDA margins are high because the business is capital intensive!
For example, here's a comparison of asset turnover between a consulting company, a SaaS company, and Center Parcs:
A hotel and leisure biz like this requires a ton of maintenance capex, along with significant interest on the debt required to finance that capex. Even if that capex is things like water slides.
So EBITDA has to be high, because a bunch of that just goes straight towards non-optional ITDA.
Some people said "this is classic vulture PE behaviour, raising prices to fleece customers."
Wrong again. Yes, Center Parcs is expensive -- they have a great product and they price it accordingly. But when you look at the free cash flow the business actually generates, it's fairly unremarkable.
In the grand scheme of PE deals, this one is fine, but not, like, a runaway success. And they're continuing to reinvest heavily in the business too.
(admittedly I didn't help by putting "price-gouged" right in the title of the article -- my British humour doesn't come throughg to the HN crowd)
Just a reminder that there's always way more nuance to every topic you can think of. Reality has a surprising amount of detail
English
@Shpigford p.s I still wouldn't have thought this would cause your homepage not to rank. I'll keep digging.
English
@Shpigford thanks! no issues there. I would definitely look at possible duplicate content caused by your pagination parameters. Might not be the issue but best practices are adding a canonical tag. More info here: sistrix.com/ask-sistrix/on…
English
Who can I pay to help debug why toolstash.com isn’t ranking for “toolstash” in Google?
English
@Shpigford Looks like Google has crawled your site - google.com/search?q=site%… But it may be an issue with duplicate content, e.g. toolstash.com/tools/automoti… and toolstash.com/tools/automoti… might look duplicated to Google. I think adding a canonical tag to the root page would help you.
English
@Shpigford That's strange. No homepage ranking for google.com/search?q=%22to…. Your sitemap and robots txt look fine. What is the status of your sitemap and number of discovered page in google search console search.google.com/search-console… ?
English