Hacker News Reader

This is exactly why I built Molecule.dev. It's brand new, using all the latest and greatest battle-tested tools.

I started it because I recognized that it took entirely too long to implement some of the most basic core functionality for cross-platform apps in a way that is up to my standards (i.e., minimal with no real vendor lock-in). It actually boggles my mind that nothing of the same quality as Molecule.dev exists yet. It's obviously a problem that developers everywhere run into when building custom apps. I've seen a few solutions, but (again) they're all either not up to my standards or they try to lock you in.

I think most developers prefer to implement things themselves, as they have their own preferences and may (like me) not trust that the code quality is up to their standards, so the difficult part here may actually be to convince developers that the code is high quality and something they themselves would write.

Interesting. If I did my math right, you have 917,294,284,800 combinations of configuration available, not counting the "other" option in each category.

That's a lot! Most are coming from things like the deployment OS (720 options since you aren't limited to only one choice).

I've never considered all the different combinations, but it is definitely quite a lot of work, and I'm planning on adding more.

My goal with this is to help developers and teams of any level, from beginners to experts, quickly get started building the highest quality product(s) they possibly can. I want to raise the bar and eventually set a new standard of DX and UX for both developers and end users.

How can something be both state-of-the-art and battle tested? I thought battle tested meant something more like "tried and true" (AKA not new).

Cool product though!

Most of the tools have been around for quite some time (on the web tech timescale, at least), and they've each evolved and improved as they've been battle-tested. For example, React is pretty state-of-the-art as it has changed (improved) significantly over the years, while still being used in production everywhere.

What you've built here is GORGEOUS and is exactly what I'm imagining the future of development will look like. It's a dream.

Until you get to the end and it says "$400" -- then I remember it's still 2022 and we're not there yet. You deserve to be compensated for this work, absolutely. I just look forward to a future where tools like this are no longer privately held.

Thanks! And yeah, I wish I could release this for free but it requires so much time and effort to do it right that I would end up homeless.

It's the right choice! When an open version comes, it will most likely be a community effort. I would do the same thing in your shoes.

I haven't tried it yet, but if it does what you say it does, I honestly think you underpriced this.

You could be right, but at the same time, many people seem to expect this for free. Perhaps they underestimate the amount of work required to implement quality core functionality, or maybe they feel that all core software should be free and open source.

I recently lowered the price in hopes that more people would be willing to follow through on buying. There have been many submissions but no one seems willing to pay. I may increase the prices again later on, but it really depends on how things go.

Ultimately, I want this to be accessible to the average developer so more people have the opportunity to quickly build awesome things and bring their ideas to life. It may prove difficult to find the optimal price point.

There's still a lot of work to done, especially on the marketing/convincing side of things. Maybe I'll look for funding, but I would prefer to bootstrap it. I'll probably need to find contract work soon if no buyers follow through, however. Ideally, clients would submit via assemble.molecule.dev and we move forward from there.

Oh man, I hate to hear that you are struggling to find traction since I think you're building the future here. I really hope you find a way to make this work so that you can be a trailblazer on this front. The sooner these types of things become popular, the sooner there will be competition and more contributions to this space, and it will become more accessible.

>There have been many submissions but no one seems willing to pay.

Tailwind is a masterclass with regard to how to market this kind of thing.

1. Release the core of your product for free (basicially give 95% away for nothing)

2. Sell premium features/components (the last 5% of your product for serious people with a budget)

Tailwind is selling their CSS framework UI component kit for $400!!!

It's a freaking simple little design system which is nothing in comparison to what you've built here and they are *rolling* in money.

Good luck.

I will probably do something like this. I really appreciate your suggestion(s) and feedback.

Second this. This product sounds amazing. But hard to believe it without seeing it. What can you charge for that enterprises/funded entities would pay for, but side projects wouldn’t need until they start making money. Like: Electron, Oracle/SQL Server support, Exchange integration, etc. Good luck! As someone who has been “working” on a SaaS project for a year, would love to try this. But can’t shell out $400 out of pocket for something I can’t touch.

Better business model might be make some api, or layer that you control in all this...basically it becomes like a headless cms...could be as simple as centralized auth/data-hosting, then the other version for $400+ could be on-prem/self-hosted...

Then people might get 1 month trial, and then $19/month/app.

This way you start building up a lot of MRR, and can maybe add price points for # of average monthly users or something like Auth0 has.

This is a really solid idea. I'll have to think about this. It could pair well with ux-app's (sibling's) suggestion.

A suggestion: maybe you should do a document/video explaining what your app does for people like me, because while I was reading I got curious, but I don't really know what it does. After spending 18 months developing my SaaS science algo in Python (self-taught) I was like: ok, now it shouldn't be too hard to make this available in a website. I only knew very basic css,js(jq) and html, so I started to learn VueJS: it was so freaking hard I realized I wouldn't be able to do it in the timeframe I had so we have decided to launch our product with an animated wireframe instead. At that point I also knew there would be challenges in 2 other areas in which I wouldn't have the necessary time/expertise: cloud/server and security.

how can you be 'latest' and 'battle-tested' at the same time?

A good example is probably React. I would consider it to be battle-tested, but it's evolved significantly over the years. "Latest and greatest" would refer to e.g. hooks in the particular case of React.

I honestly feel like you have way too many options. I went to select for some imaginary app.

I do appreciate your work, don't get me wrong, but maybe iteration couldn't be bad to happen. Talk to few people, see what they say.

I agree! There is a lot involved in building full-stack apps, especially with all the different tools available, and Molecule.dev barely scratches the surface.

I'll soon be adding presets and templates, which should reduce the complexity of selection. People often have requirements which I have not considered though, so the ability to customize everything will likely remain for them.

Counterpoint, I didn't think there were too many options. It took me <1m to scroll through the whole list and select what I wanted, ignoring the ones I wasn't interested in. I thought it was perfect. If it was 5m, on the other hand, or had page after page of options with no end in sight, that would become tedious I think. But this was very brief and straightforward.

This has to be the coolest experience I've ever had with a landing page

That is very kind of you to say. I worked hard on it, but I think it can be better. Marketing and succinctly communicating the value of a solution to a problem has never been my strong suit.

I'd AB test a different color scheme. I don't mind but I know some managerial types that would.

When I click the “get started” link and go to https://assemble.molecule.dev I get a blank page on iOS safari.

Looks like Safari was being extra strict about the CSP (content security policy). It should load on Safari now. Thank you!

This is amazing.

Could you elaborate on the "No vendor lock-in" part? When we select cloud services like Heroku or AWS, aren't we technically locked-in? For example, AWS Amplify will create DNS entries and its own SSL certificate for the domain. If we wanted to switch, we'll have to set these up again in the new provider.

How does Molecule not make you locked-in to the providers you select?

i take it to mean you're not locked in to molecule directly.

Yes, this is also the case. (Referencing my sibling reply.)

Great question. As it is now, Molecules are designed to be as independent (platform/vendor agnostic) and flexible as possible. For example, when I was trying out all the different cloud services myself, I was able to connect each of them to my git repository, and they would each build and deploy the same exact codebase with no extra setup beyond defining a few environment variables. It took zero effort to switch from one service to another. I take no credit for this though, as it's the result of these services being designed so well, competing with each other. I only ensure that it works perfectly for your selection and provide instructions to help you get started as quickly as possible.

As for other types of services, like cross-platform payment processing, push notifications, certain libraries/frameworks, etc., some of these try to lock you in, which are all avoided. Your Molecule's code covers all of this on its own so you don't have to rely on a third party. But at the same time, sometimes it can be cost effective to use a third party service, and in which case, the code is written so that adding and removing features and integrations of your own is super easy. I try to make no assumptions about what you will eventually need to add (or remove), while providing the most solid foundation possible for you.

Like others, this sounds amazing and I'd totally pay $400 for it if I could try it out somehow first. Some kind of freememium model makes more sense to me. I'm an experienced dev with plenty of my own ideas, but I'm not experienced in everything and I always balk at how much drudge work I'd need to do to really get something going. This sounds awesome and super valuable.

Just some feedback! Because of your message here I read through, but just the intro and headline would have been too abstract for me and didn't solve anythings specific for me:

"Assemble a simple but powerful, fully functional codebase consisting of only what you need."

Maybe add a small list of features you supply out of the box? Auth, Subscription management, billing, etc...?

You've done a great job here. I've been a PM in this space for 10+ years at major cloud providers. DM me if you want to brainstorm on GTM / monetization levers.

> It is infinitely easier to build and deploy things in 2020 than it was in 2010, and I'm hoping the time between now and 2030 represents a similar jump.

I'm not convinced that is true. At least if you compare 2012 with 2022, it was very easy to get a Rails app working, with user authentication/signup/password reset/etc (Devise), integrate Stripe for payments, deploy the whole thing on Heroku. It wasn't an overnight project, but one person could get that done in a few weeks of full time work. There was a lot of energy and excitement around this stack and the community often created "one good way" of doing things.

In 2022, all those technologies are still available, but they are uncool and most of the internet will dissuade you from using them. Instead, unopinionated micro-frameworks are in vogue, and you you're down a rabbit hole creating the same product by gluing together dozens of libraries, writing a jamstack SPA, deploying the whole thing on kubernetes. You could easily spend several weeks just building out user signup flows, password reset, email templates, etc.

I'm not saying it's all bad today: you can do much more with a SPA than you ever could with server side HTML + jquery, and back in the day when you hit a wall with Rails it was often extremely difficult to work around, whereas today's libraries are much more flexible. But for the happy path of "scaffold a functioning SaaS app", I definitely feel like it's harder to get started today than 10 years ago.

> At least if you compare 2012 with 2022, it was very easy to get a Rails app working, with user authentication/signup/password reset/etc (Devise), integrate Stripe for payments, deploy the whole thing on Heroku. It wasn't an overnight project, but one person could get that done in a few weeks of full time work.

If you use Phoenix in 2022, you can generate authentication/signup/pasword reset/etc with a single command and the generator was written by the same person who created Devise back in 2010. You also get LiveView (and generators for it) that let you make apps with all the interactivity of a SPA.

Oh, and there's a single line mix command to deploy a release of your app to fly.io!

In short, things have become far easier for small teams and solo devs as long as they chose appropriate tools. Obviously it's going to be painful if you try to go it alone with the exact same tools a team of 30 or more would use. The same was just as true a decade ago.

The issue with LiveView and similar solutions like HotWire is latency.

That's why Hey is impossible to use from places like NZ.

I haven’t tried Phoenix, but I believe you. I guess my point is more that the “general internet consensus” in 2022 doesn’t make life easy for you. It might be possible to do all this easily in Phoenix, but most people starting out are going to get directed down the npm/js/nextjs/express/etc DIY route, which is probably not the easiest path to market.

I mean, it is years of work, speaking from experience. I make a similar tool (but for Node.js/Vue, if you're not a fan of Django), and it took around two years to get it to where it is, with user account creation/management, subscriptions, teams, admin dashboard, scripting, etc. I'm definitely not charging as much as Bullet Train, but also I feel fully justified in charging instead of open sourcing it. It's been my sole serious side-project for years now after all.

The problem for trials of stuff like this, is that you have no real way of enforcing it. You just have to hope that trialers feel like doing the right thing and paying if they continue to use it, because trying to recover a few hundred dollars through legal means is just not worth the time and money you'd have to spend.

In the end, I settled on a 60-day money-back guarantee, and I've already issued a few already, to folks who thought it had a feature it didn't yet, or ended up not pursuing their idea, etc, so it seems to work pretty well.

You missed a great opportunity to plug your product on HN, give us a name and a link :)

nodewood.com

I didn't build this app on it, but if I did it again I would probably use Jumpstart. https://jumpstartrails.com

There's also an open source version that handles a good chunk of the stuff needed to get a Rails saas up and running.

yeah I was going to plug that too...

the open source one handles a basic devise, admin backend and some other things. the pay one adds stripe and payments and stuff like that. but the open source one is really nice for hitting the ground running

It does take a lot of time and experience to create these types of solutions.

For example I put together https://buildasaasappwithflask.com/. It's a video course and SAAS example app that ties together a bunch of features. I've easily put in over 1,000 hours to create and update it over time. It's also been extracted from building multiple SAAS apps for clients (I do contract work).

The way I look at it, it's paying $59 once to get a well tested code base that's going to save you 100+ hours of research and development. I also give lifetime updates and support along with refunds for up to 1 year. For context since I created it in 2015 I've re-recorded the course twice and continuously add a bunch of incremental updates. Every few months I go in and update things to their latest versions. Answered a countless number of questions too.

I think the key is having at least something done. Nowadays, most of my new projects are done by copy pasting things I've done in previous ones. This reduces the time spent setting things up considerably.

Laravel is great in this regard. They have a number of first party packages like Laravel Spark that handle the heavy lifting of subscriptions, Laravel Nova that lets you spin up admin panels real fast, Forge for hosting... Highly recommend checking it out if you don’t mind PHP.

how do it compare to django? my experience with php was limited to some wordpress hacking.

Laravel is pretty good. Spark definitely reminds me a lot of Django. https://spark.laravel.com/

I think nowadays you can use whatever frontend you want with it, where prior versions you had to use Vue (I think)

I haven't used Django so I can't comment about that. However, the latest Laravel Spark is now a totally separate part of your app that you don't control. It's super easy to implement, include the package, configure your packages and be done with that part of it. Then within your application you have a bunch of methods available on your billable object.

I'm really not doing it much justice, but the docs are pretty short and easy to read through: https://spark.laravel.com/docs/1.x/spark-stripe/plans.html

What's even harder is product-market fit.

I can blitz through building a basic SaaS product, but will it be worthwhile to:

1. Make enough money to replace $dayjob.

2. Make enough money to even cover the headache of supporting customers.

3. Make enough money to even cover the hosting cost.

As someone who has built and shipped lots of projects, both for work/startups as well as personal/side ones, I would say that most of the time with building/shipping products quickly, tons of corners are cut. It may not be obvious at first sight but you could see if you look for it. That's typically how one person can build and ship something in very short time. As a simple example, in HN, tech circles and the software engineering industry we talk about node, react, microservices, architecture design, etc. but when you're one person trying to ship a product, there's a high chance you're writing some basic HTML, jquery and some php backend, rather than a react SPA with SSR/SSG. (Not saying this is how OP did it, just in my experience how I and a lot of others did it)

You're also right that it's infinitely easier to build and deploy things in 2020 than it was in 2010!

> I find it daunting merely to design a landing page with user sign-up, let alone a feature worth paying for and an account subscription flow. There's just so much to do. It seems like years of work.

All of those things are built into Jumpstart Rails, which the author now uses. To be 100% honest, just using a highly productive framework like Rails, Laravel or, my personal favorite, Phoenix, goes a long ways towards making it quick to build all of those features. If you're experienced with a Rails-inspired fullstack framework, I'd say the work involved is closer to a week or two at most.

Jumpstart gives you a sane default for that week or two of work.

It isn't! I started without any development experience and built an MVP in 3 weeks with user sign up, pricing and subscription using some getting started guides and stakoverflow. Not best practise or anything, I'm sure you laugh if you saw some of the code. Taking card payments was something I was worried about but Stripe has pretty good documentation for a layperson.