Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: Sourcery – get a free, self-updating website (sourceryapp.com)
230 points by jarnoux 25 days ago | hide | past | web | favorite | 73 comments

Landing page is crisp and precise. I love the idea. Congratulations on the launch.

Auto-complete is fast and relevant. I believe you're boosting local results? That's super nice. The generated webpage is nicely filled with details from Google Maps (?) along with reviews and a contact-us form. I like the overall simplicity of adding and editing subsections of the pages, too. Kudos.

Quick questions:

1. From where'd you scrape data for businesses world-wide, if you don't mind revealing? The data you have (judging by auto-complete) is frankly unsettling but not surprising.

2. There's e-commerce too, and the flow is manual: Upload photos, add price and description. Simple enough. As your product evolves, how'd you plan to manage to keep it that way rather than turn it into a Shopify-clone which most business owners would then need a IT guy to run it for them...?

3. On the pricing page, you mention business plans are coming soon, but the only two features I found that's extra from the free plan are 24/7 support and ...and more. Curious: what more?

4. How long have you been working on this idea? What were you inspired by, initially?

5. How do you generate the pages? Randomly choose a template and fit the data in, or is it powered by AI? I see that no matter what I choose as a Goal among Brand, Leads, and Conversion the template and the structure of the page remains the same.

6. Can you give us a 10000-foot view of your architecture and tools, if that's okay?

7. When you mention personalize the website for each visitor... Is that like ad networks that show relevant ads where you mean to show relevant products? If so, how'd you gather enough data abt a visitor to be sure of the personalization (acquire/gather data from third-parties)? What value do you see that personalization offers? I'm not being skeptical, just want to hear your point of view on this.


Thanks for the good questions!

1. Mostly google maps and social networks

2. At the moment we don't have a "back-office" integration, we have a rudimentary editor that can be manually edited at the back of our "Products & Services" feature and if the user provides their PayPal merchant ID we add "Purchase" and "Subscribe" and "Pay Invoice" buttons wherever it makes sense.

3. We have a number of ideas for that, including specialized business platform integrations, custom domain names, and other features that would be used by businesses that we think should be on a paid plan.

4. We've been working on this for a few months now, we were inspired by the readily available and fresh data and the price (in time and/or $$) to get a decent website created and updated for a small shop people like restaurants, barbers, nannies, garages, etc.

5. Given your business category, location, amount of data available and other signals we try to be smart about what your site should look like and what sections should be where. That stuff has been figured out with A/B testing every day for the past 20 years and humans should move on. That also means we can generate the website in real-time, for the specific business and the individual visitor making the request...

6. Heroku + EC2, mongo atlas for the DB and lots or nodejs and react. letsencrypt for the certs.

7. It means we're able to shape the site depending on the history of the visitor. For example we'll show a menu first to a visitor who is likely hungry.

> For example we'll show a menu first to a visitor who is likely hungry.

How would you possibly know this?

If I go to a restaurant's website I want to know two, maybe there things, their hours, the menu, and possibly their phone number/reservation link. I have never wanted anything else.

As simple as it sound, I feel there's need to stress on how important it is for users to find what they seeks without going through maze of hyperlinks and visual fluffs.

* If I'm seeking for contact number and/or address, I'll attempt to find it in "Contact Us" on first level navigation, or footer of any page.

* If I'm seeking for menu, I wish it's available on first level navigation. Price would be nice, but understandable if it's not feasible/ideal to show them.

* If I want to make online booking, I either look for widget on side of homepage, or proceed to "Contact Us" to see if there's either a widget or hyperlink for booking. If not, then at least I should see a telephone number for me to call.

* If I want to see more details of an event I heard of that is happening now/soon, then I wish it's available as a feature on homepage. If not, I'll look for "News" on first level navigation and hoping to see what I'm after in the first page of listing. And for the love of god, please include year in your news post or in the body content. This is especially important when it is an annual/reoccurring event.

I’m guessing visits around 12PM and around 7PM

So people out of those hours looking to book something for the evening are not interested in the menu?

I thought office 2003 learned us just how miserable these optimizations are, a consistent experience is always better.

1. Google Places API.

While the concept is sound, I do hope that you will allow users of your service to 'detach' at some point to make their sites into something more complex. Working around Website as a Service limitations is a tremendously tedious haul. I have worked with a Shopify project for a few months now and having to either write or pay for (via an app) functionality that has existed on any self hosted e-commerce for years is really annoying. Not being able to modify backend code impedes development speed as you try to hack around the backend's deficiencies with Liquid. I didn't want this post to be this long but I just wanted to give my experience on this kind of subject and I hope you take appropiate measures to save your customers from dealing with similar issues.

If you don't want to pay for the Shopify app you can always write it yourself..

I am already aware of that, but why should I have to for basic features? If you need reviews for products, you will need an app developed by Shopify (but for some reason not integrated into their main product - what the hell?) which doesn't interoperate properly with any multi-language app (again, which shouldn't even be an app, as multi-language is a core part of many e-commerce applications) since it generates HTML server-side with embedded texts and returns it as JSONP. Another example is getting orders in a printable format. You need an app for something that could probably be done in 100 lines or less on the Shopify admin's backend. Don't get me started on multi-currency support which you can't even do with an app and need to rewire your entire checkout process to a third party system for. It's an agonizing mess.


Sorry, what? Are you implying that companies I write code for go out of business? Why do you think that? >well-maintained application architecture Tsk tsk. >WooCommerce Hell I have never experienced this. Can you elaborate?

That dude seems unnaturally defensive of shopify for some reason...I'd just ignore it and move on

No, I can't. Sorry. Maybe I can, if you disclose what companies you are writing code for and not just plain namedropping, but disclaim the actual products you are involved with.

I thought the concept is "ai" websites, why do you write it is sound?

"Sound" in this context means "makes sense".

Please change `width: 100vw` to `width: 100%` in the header element on your landing page. It results in the horizontal scroll bar in the bottom of the screen (you will see it if you view your landing page on something other than a mac).


Since Google Places API doesn't allow storing any data, are you just footing the bill for the site to be "re-created" every time a user visits it?

You're offering this for free, which is great, but I'm not sure how this works out financially.

You can easily use Web Scrapping using Javascript dataextraction and storing into a CSV.

It’s slightly sad that the internet is turning into a bunch of sites that are all based on the same templates. Don’t we need more website builders that are open and creative by design? I understand not everyone has the time to learn how to write html but we should at least be building tools to make websites more interesting and distinct.

I second your sentiment. I ask myself, if at least one of the reasons behind this is the human need of recognition.

I mean don't we all want our presented thoughts to be seen/read if not appreciated?

But how do if your quirky design is being punished by the search engine overlords, because it doesn't conform to the standards defined by them? By their rules leading to SEO spam and mediocre, conform designs?

Just an idea not totally thought through.

You want Neocities.org, the spiritual successor to 90s-era website quirkiness.

How do you verify that it is exactly business owner who's trying to create website? Ease of generation almost invites pranks, and maybe even fraud. IANAL, but I strongly suspect you'll got sued at some point for damages caused by a fake website.

In their defense, anyone could buy a domain name of any unclaimed business' today and setup a website and email accounts.

SourceryApp does pre-populate the phone number attained from Google Maps for auth but that remains editable, currently. I believe the problem of fake websites will bite them if they get anywhere as popular as Twitter, but they don't have to worry abt that, right now, imo.

From the terms of service... “ In other words, you grant us and our subsidiaries affiliates, and successors a worldwide, non-exclusive, royalty-free, fully-paid, transferable, irrevocable, perpetual, and sub-licensable right to use, reproduce, modify, adapt, publish, prepare derivative works of, distribute, publicly perform, and publicly display your Content throughout the world in any media.” lol nope

Thanks for the remark. This simply makes sure we have your permission to use the content you send us to generate and host the website you're asking us to create.

Is the "irrevocable, perpetual, and sub-licensable" really necessary for this?

And how would you expect them to host your website content with the right to publish it?

soooo.... same terms as Google Apps then.

Yeah but they ain’t Google...

In fact the Register for the domain is Domainsbyproxy, that crap that was being used during the last election to hide sites harvesting data....

The registrar for the domain is likely GoDaddy.

There is absolutely nothing special about Domains by Proxy.

If you place your domain with eg Cloudflare, they'll conceal your registration information. Dozens of other prominent registrars will do the same thing for either free and inexpensively.

I think this is pretty brilliant. I offer a word of warning: SMBs are a tough crowd to sell to. Even with a 100% turnkey offering, they are a remarkably vocal, high-maintenance group as a whole. The idea of the external web, the social web as a CMS has lots of implications in enterprise as well. Don't let SMBs sour your love for your business.

Very slick! Tried it for a no-online-presence second hand retro gaming store here in Miami and it spit out a decent website!


Pretty damn good lads! Congratulations on the launch!

This is a pretty solid, simple idea, gotta say. Of all the "quick website" apps, this approach seems like it could get a pretty good result from the get-go, given a good amount of social media / website information available for the business.

Curious, what other quick website apps you know of? Not low code, but quick.

I've seen https://milkshake.app, https://onuniverse.com before and walked away impressed. https://glideapps.com is in a similar domain, but for apps.

The most minimalist that I'm aware of: http://txti.es/

Thanks for that, I had the same question mostly because their site is down.. It's to bad you gotta install an app to try or even see the basics of how both of these sites work.

Strikingly is one

The problem is that the business owner already uses Facebook to do marketing, as it's much more effective than just a website.

I got this error message while trying to generate a website:

{"message":"Cannot return null for non-nullable field Entity.phone."}

happens in the rare cases when we can't find a phone anywhere for the business - we're working on a solution.

I got this too.

My business doesn’t come up in the autocomplete and there’s no category for the type of business. There’s no free text field for me to enter my business type and if I choose fixed categories from the list, none of the imagery in the generated sites looks passable.

I don’t understand what functionality this offers that GoDaddy GoCentral doesn’t have? Why do I care if it happens in 5 seconds if the result doesn’t feel tailored to me or quality?

I can’t help but feel like it’s a bargain basement ripoff of a hodge podge of ideas already in the market from Squarespace and GoCentral.

That's fair, Squarespace has a use case. Sourcery is opinionated and not a freestyle tool - we are building websites that you don't have to design, build or update and that lets us offer features that aren't possible in DIY site builders.

I think GoCentral already does this: autogenerated single page sites where the imagery & text are dynamically generated based on a small number of user-made choices at sign-up.

I tried a couple of companies but I just get a blank page and a bunch of JS errors in the console :/

should be fixed now

we're hitting API limits and themes are defaulting to generic - sorry about that, we're working on it

ok, themes are back in business

Just a head's up to the site admin that browsers may disable autoplay on video, resulting in a blank box next to the "How it works" paragraphs.

You can enable autoplay by including a `mute` attribute in the <video> tag.

thanks for the advice, `muted` was added

How do you proceed if your business isn't found in the autocomplete?

I tried it. It still works, looks like you’ll have to provide more info though

Tried a few restaurants in my area.

RT Rotisserie in San Francisco yields an error: {"message":"Cannot return null for non-nullable field Entity.phone."}

Add a <noscript> to your landing mayhaps? Takes a little while to figure out that I need my JS on for a proper representation.

great project. But why? Do you intend to out-compete wix/etc?

This is a significant piece of work, effort, and skill. I feel so bewildered some times. Maybe i am just too focused on end-result and revenues.

p.s. yes, i am working on my own startup. yes, i am a developer not a manager. yes, i do get sidetracked into side projects

p.s.s. i love the pay-pal integration

thanks for the kind words

Oops... we've encountered an error. GraphQL error: Cannot read property 'alias' of undefined GraphQL error: Cannot read property 'alias' of undefined

My site doesn't look as good as I was expecting: https://i.imgur.com/6c89cqu.png

should be fixed - does it look better now? :)

Nice. I wanted to do this when I was at Weebly, but never happened.

So basically you're taking all the content under a Google Business Listing and flattening it out. It's slick but nothing crazy.

Oops... we've encountered an error. GraphQL error: Cannot return null for non-nullable field Entity.phone.

Please check the designs on Firefox on Android. Doesn't work properly.

Okay, I'm seriously impressed, this is amazing! Congrats on launching.

Really cool idea

typed: angryangels, got Angry chicken xD


Hey, can you please not break the site rules like this? We're trying for a bit better kind of internet discussion here.



I can but I won't.

Who are you and why would you think I'd trust my business to you?

>trust my business to you?

Clearly you haven't taken a proper look. They are just scraping what is already there, so they aren't the ones holding the trust you mention.

Applications are open for YC Winter 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact