Show HN: SaaS application boilerplate | Hacker News Reader

Show HN: SaaS application boilerplate

github.com

107 pointsKRains74 comments

I appreciate the efforts being done in this space, but I don't think this is launch ready yet. It doesn't include some of the things that make a SaaS a SaaS, and not a typical user sign up web app with admin panel.

For a SaaS framework you usually need at least: integration with a billing provider (almost always Stripe nowadays), web hook handling for a bunch of services (or a generic way of handling webhooks easily), some kind of social auth integration framework or ready built integrations for a bunch of services, bulk email support, a modular dashboard (seems you have the start of this?), a landing page generator with pricing tiers that integrate with the backend somehow, a team management option for users.

Have a look at Laravel Spark, or some of the other premium SaaS frameworks available right now for what people a now expecting of these frameworks.

Also, I'm a little dubious about your license (seems to be fixed now?). You need to be crystal clear about whether or not this is open source, or if you are trying to sell a product. "SaaS as a Service" (or SaaSaaS as I like to call it) seems to have blown up recently, but I'm usually wary of things like this. When making a SaaS, I like to keep margins and fuss as low as possible, and including a piece of software that might cause me legal trouble down the track is a big no-no.

Thanks a lot for the great comment.

I totally agree about the functionality. Stripe integration will be the next feature I will add. And even so, I still believe this boilerplate can save you at least a week of work. I already used it for my other projects and I can confirm it.

Dashboard UI is already done and I'm going to add more modules/components to make it easier to use. Do you have any suggestions what it would be?

Laravel Spark is one of the resources for inspiration :))) I just didn't implement everything I would like to - yet.

I fixed the license, the boilerplate can be used for any purposes (except ones against the law) - this version. Because I'm working on the commercial extension of it - and it will be a completely separate story. But even with using of commercial modules (or version) you will be totally free to use it and it will be completely white-labeled.

Another situation when you are trying to modify and resell a product but it has nothing to do with making a SaaS, right?

You should really consider picking a standard open source license if you want people to try your work out. I personally wouldn’t consider building a business off of a license like that.

Yeah, I'm thinking the same. I just didn't really have the time to put some standard one - was busy with all these comments!

Okay, I put MIT license. I found this type of license is most appropriate for this product.

I agree, i'm working on my own SaaS boilerplate in laravel debating oss vs paid, might be a cheaper Nova/Spark combo alternative. $99 per project is pricy, esp when a lot of projects are just testing ideas and things.

SaaS needs to have teams, projects, payments, auth, invites, and possibly be configurable for different 'types' of scenarios... projects could be optional for instance if they don't group actions by project, invites could be optional if anyone can join any team, heck even teams can be optional if you can't add multiple orgs but it's nice to have, say you run multiple brands or something....and want separate invoices with separate logos... etc...

I'm getting close to launch, thinking I might do it as a paid $9.99/month gets unlimited access to everything I do in the scope of the project. Might raise prices eventually but grandfather people in for a time. Or might do FOSS with addons that are pricier, but I don't like apps that do that -- I'd rather pay small fee monthly and use what I need depending on client.

Thanks for sharing your thoughts and experience! Honestly, I don't know if teams and projects should be a part of SaaS. If so, in which context?

I wonder, how are you going to charge? Is your product is app or it's SaaS itself? And I think $9.99/month is very low price.

Any link to share? ;-)

Good luck!

Look at some apps like Clockify, Slack, Trello, Jira, Asana, all have teams, -- I really like the way clockify does workspaces, and then you clients, and projects per client. All organized nicely. This sort of model of teams/projects would probably be quite common in SaaS...think anything tagetting SMB's that need to give access to parts of the app for their employees or contractors, or use it to organize, etc...

Ok, thanks! I will consider adding teams.

Although there are literally dozens of tools out there that aim to facilitate SaaS, if you look at the definition of SaaS and what makes SaaS, SaaS. Things like “on demand” activation, licensing models, and resource management are the key definition points, not so much the nuts and bolts libraries needed behind the scenes.

I wrote a funcional specification for a SaaS provider platform about a month ago (see https://abstractionlayer.jamesbiv.tech/Abstraction%20Layer%2...) its initial chapters deal with the definition of SaaS.

WOW! You put a lot of work into it, and I'm going to research it. Just wondering, what made you create this phenomenal work?

Hi KRains

Thanks for the kind words. To answer your question, I can say I had a many reasons to create this document. First and foremost I’d like to create an open source project around these specifications and see if I could generate interest and invite contributors to jump on board. Second, is to provide a working document for the community scrutinize and critique as it’s still very much work in progress, and before committing to building the solution I wanted to ensure what is built is done so in a way that gains acceptance and use.

Also, I see that Cloud and more particularly SaaS has a lot of bias surrounding it’s use both from a commercial and personal perspective. This largely has to do with how SaaS providers store and use peoples’ data. My aim with PAL is to create a flexible solution that takes all perspectives into consideration based on the type of offering and what is acceptable by the customer or user of the SaaS offering.

I could go on further to say that since SaaS has become such a large segment of the Cloud market I also feel it’s important to keep as much of it as open and free as possible as to allow smaller players the chance to offer SaaS solutions. Since the market is largely dominated by Big Tech this is becoming increasingly more difficult and that has a lot to do with the lack of resources available to smaller companies. It may not seem apparent to most people especially those in countries where internet services are easily available but currently where I’m located, a country where the economy suffers greatly, the low cost of establishing SaaS solutions for businesses vs offering custom built in-house solutions is a far more feasible option.

Anyway, I hope the you find the contents of the document useful and helps you in some way with your project :)

Regards James

Okay, I got your position. And thanks a lot for your efforts! I will contact you if I have questions/suggestions (I found your email at the end of the doc).

I am a fan of opinionated full-stack frameworks, in particular for building B2B web applications which contain lots of CRUDL views.

For an open-source solution that largely revolves around the React frontend, consider react-admin [1] + its adapter for Hasura [2].

[1]: https://github.com/marmelab/react-admin [2]: https://github.com/hasura/ra-data-hasura, https://github.com/hasura/graphql-engine/

In Ruby/Rails-land, there is https://bullettrain.co.

Anything full stack with Python/PostgreSQL/Angular?

I've previously found this site helpful for a high-level view of the ecosystem: https://www.fullstackpython.com/angular.html

Nice thanks - I had not been here. The first two tutorial links are pretty old there are to AngularJS.

Thanks for the link - pretty interesting.

Maybe you would think to move to ReactJS from Angular?

Yes I'll still take a look. I subscribed to your list and Twitter, kind of interested in what you are up to.

THANKS!!

Not Angular, but there's https://www.saaspegasus.com/ which is bulit on top of Django.

Full disclosure: I'm the creator. Happy to answer any questions you have about it!

Thanks, will look at them.

Big fan of BulletTrain - it's a big accelerant for building a SAAS (literally worth a couple of months of development time).

I saw them, read their blog (for inspirational purposes of course :) and even wrote them. But I believe it's really overpriced. I'm working on the paid modules/components but I will keep the price moderate.

FWIW - I found it valuable as it provided both the sort of modules/components as you're describing but also a set of productivity accelerating things like the SuperScaffolding etc to really apply those into an application along with a well thought out system to extend all of that when I hit the limits.

Scaffolding and other cool things for productivity accelerating are in our plans.

Check some of them on our website https://www.saasforge.dev in the "Coming soon" section.

Hey KRains, author of Bullet Train here! Congratulations on your launch and making the front page of HN! I know how great a feeling that is after all your hard work!

I disagree that Bullet Train is "really overpriced". The features it provides to customers can save them hundreds of hours of development time in total. There are individual features in Bullet Train (the teams functionality, the Stripe integration, outbound webhooks, scaffolding real-time chat conversation threads onto models and an inbox for all of them, Zapier integration, just to name a few) that would each save most development teams at least $1,450 in development costs, and that's before you take into account the power of Super Scaffolding for code generation.

Bullet Train will save most development teams months in effort right off the bat. That's well worth $1,450 to the customers who have been able to get to market faster with components they know are being used and maintained across many different products. Bullet Train is only two years old, and there are already folks running seven-figure businesses around products that cost five-figures to build using Bullet Train. I'm not trying to imply it's magic, I'm just trying to point out that there are situations where $1,450 can be a no-brainer.

I love that you're starting with an open source base. If I knew when I started what I now know after two years of selling Bullet Train, I would probably do something similar. But you will need (as I've needed and as others who are doing similar projects have needed) help along the way. (For example, you'll definitely want to get the support of a designer.) Unless you want to bring those people on as partners, that help will cost you money, sometimes lots of it, so make sure you're charging enough for your premium components to pay for the help you need to make the product a success and the best it can be.

Best of luck! Wishing you a lot of success!

there's another guy on here saying $99 for Laravel Spark is pricey

per project btw

Hi Andrew!

Thanks for your great comment, really appreciate it!

You are right, I'm very excited on this experience! Even if it's not the first time when I publish my products here, I never had such level of attention.

Well, I'm looking right now from my personal point of view and your price looks "a little bit" high to me. I know I can do almost everything you already have in your project and maybe more, and if I would like to create my own SaaS right now, I wouldn't pay so much.

But if considers only corporate customers like established companies or well-funded startup, of course, your price is just nothing. The problem is I'm still looking for my niche, ideal customers etc. and I see this price is high for many of them.

My goal is to provide a democratic price that would be moderate for individuals and companies but would be different for them. I'm thinking to charge per seat as an individual can buy just one license and it will not be hard for him/her and companies would pay much more just because they would have more seats. But it's still plans (yet).

Thanks again for your advice and comment, good luck to you too!

Wooooow.

You better be super sure you can launch something good in 1 year with a $1,500 / year license.

Having "Open Source" in the name is slightly misleading...

> License

Thanks! I have to fix this sentence...

Fixed! But I still left the part about not using the code for purposes prohibited by law...

It would be interesting to see a court case in which the offender violates this clause of the license.

Haha! But at least nobody would sue me for permitting improper using of it :-0

It's (presumably) redundant for what you intend it for (you can't grant a right to break the law..) - but can still be troubling. Which law(s), which jurisdictions ?

Copyright law is fairly international through trade agreements - which is one of the reasons copyleft was such a great hack.

I changed all those words to just MIT :)

Looks great! I also recently worked on a similar project, Staart (https://staart.js.org) but with a Node.js and Vue.js stack using TypeScript.

It has typical SaaS features like onboarding, API gateway, social auth, Stripe recurring billing, teams, etc.

Thanks for sharing your link. It looks pretty mature to me! But I see you put all parts separately in Github, is it more convenient?

Reminds me of https://spark.laravel.com/

We all are part of the family called "boilerplates" :))))

Nice effort but I don't like it.

You could replace everything here with Django. There's nothing in here that supports migrations, deployment etc (Django can).

I've used React and Webpack and loath it, it's extremely opinionated, saturated with pure JS developers, has an enormous learning curve and load of gotchas.

Django also has API support and a wealth of add-ons for services like federated auth, ElasticSearch bindings, bindings for S3, Redis, discovery, Cassandra, NoSql, etc etc.

The architecture here is also really unscalable. There's nothing in here that can be horizontally scaled, and Jinja or Django + bootstrap is 10x faster for prototyping before you hire a dedicated dev to maintain a React nightmare.

Modules are not a feature of this app, they're a feature of the Python language. Putting that in the README is misleading IMO.

Thanks for your opinion! It's probably more valuable for me than from people who like it :)

But you are a little bit wrong. Currently, it supports migrations and will support deployment very soon.

I don't agree about ReactJS and Webpack. I start using both in my projects and it didn't take too much time to learn them. Even if I don't know them deep I know them enough to use in the real projects.

I agree that currently, Django is more might than this project but it's just a beginning.

You are not right about modules too. It has nothing to do with Python's modules, I "invented" my own "modules" - it's just a combination of front-end code and Python API in one folder. And I created them to make the project more scalable. I don't know though if I gained this goal or not - yet.

I have briefly evaluated a few of these in the last month (focusing only on TS or Python) or so and have ended up convincing myself I need to go down the path of building bits I need in my own way. Consider this only constrictive criticism and I massively applaud the hard work that has gone into these projects:

   https://github.com/async-labs/saas

Good:

-React Material UI Front End

-Book showing how you how to build it, really cool way of funding OSS!

-Simple but scalable enough architecture, not too much cruft.

Not for me:

-MongoDB/Mongoose, where SQL (sequalize ORM or plain postgres) would have been a much better fit for a SAAS boilerplate in my opinion. Quite a big user document in the mongodb schema.

-Quite tight integration with AWS by default (s3,SES,Lambda) not hard to replace but the principle of OSS with defaults to one vendors IAAS feels uncomfortable to me.

-No password based login, not a big deal but an interesting choice.

-No Delete on team members and some other simple things that are needed to make it functional.

-Missing a few key features (permissions, master dashboard ect) that are addons by commercial licence which is fine, but these add up quickly to ~$1000 for something that would still be a fairly basic boilerplate.

   https://github.com/staart/api

Good:

-Awesome features: landing page, RBAC (although roles are hardcoded so take a look at these before you build your app), TFA, Api Gateway, Auto Join company based on domain. Everything you need (and some you wont)

Not for me:

-While there are lots of features, some don't actually work out of the box and require fixes. Some features you will not need and may have to strip out. If more time was spent on getting fewer core features absolutely perfect this would be the best SAAS starter out there in my view (in TS/JS).

-Again integration with AWS by default (SES) with no SMTP or second option, easy to replace though.

- Would personally have chose postgres with knex (rather than direct mysql) for the DB and react for the front end over vue but these are personal preferences.

- UX on front end could do with a little work.

SAASForge seems interesting but missing alot of features like companies/teams master dashboard ect.

Thanks for your great comment.

You are not the first person who wants companies/teams. But I'm not sure I understand you well, guys. Can you give more details, with some examples, why and what it could be useful for?

Ha! This is awesome! I did something similar in 2016, and called it something silly like SaaS-in-Box, but after integrating security, and Blueprints and multi-tenancy and caching, and ... (I do think I ran into some problem with blueprints, actually), ... I finally decided I was inventing Django-on-Flask and that it probably didn't solve as many problems as I originally thought it would.

BUT! That said - I think there is a solution gap here. Keep it up!

Thanks! It's a pity you gave it up, maybe you could create something really cool.

> If you experience such errors:

> In the database remove the record from alembic_version table

> Remove any files from your computer under app/migrations.

I'm not sure advising users to delete all their migrations is the right thing to do (which is what this appears to be doing).

To be honest me either. The problem with these files is if you keep them you will not able to update the database because your last file has one version and database does another. You could just remove the last migration file, that would work too.

As for the problem overall, I didn't find a good solution for it. Maybe you can help?

Nice! I have a similar boilerplate that I've built for myself (there's no client-side), Flask with JWT, login, logout, etc. The React UI looks really nice here.

Thanks! Do you have your boilerplate on Github or anywhere else?

No, I haven't felt like it's really ready for anyone else to use. Just something I use as a shortcut when starting new projects. I ought to go through it and clean it up and put it out there though.

Yeah, I understand you. But you know what - I found if you would like to share your work with people in any way (open-source, commercial or whatever) it's much better to show it as soon as possible even if you think it's not ready yet.

Quote: If You're Not Embarrassed By The First Version Of Your Product, You’ve Launched Too Late :))) (Reid Hoffman, founder of Linkedin)

This is great. Thank you so much!!

One request is to include some recommended production deployment. I have spent way too much time on devops every time I start on my weekend project.

Thanks! You know, deployment automation is one of my highest priority tasks. I even wrote a book on it (https://www.saasforge.dev/books/the-book-on-deployment) but I think it would be much better to add some scripts/automation into the boilerplate.

thanks for the link. I long for the day where I can single click and have Flask/ReactJS boilerplate setup with functional public www and local dev environment setup. I only want to focus on business logic. Sorta like Heroku on steroid? (read: I would pay for that)

Thanks for your idea! Btw I created a simple script that literally does the same - you just provide comments and it's sending to GitHub, rebuild the frontend and send command to deploy to AWS. But it's when I already set up everything. I'm thinking to automate much more - the WHOLE process of deployment, from the beginning, and actually, I already did it partially (you can find working snippets in the book). So, stay tuned to know when I implement it :)

I remember when there was a Rails one... It worked great, until it didn't. It was, however, a great learning tool on how you could role your own Rails SaaS.

What happened to it? Was it open-source or commercial?

Nice, but I think you need also an ORM. I used Flask with ORM from Django after I tested SQLAlchemy, and now I think that was a good choice.

ORM is used in this boilerplate, and it's SQLAlchemy. I need to mention it somewhere.

I am working on the same idea, on a different stack. Hope to finish it soon. Good luck with your project!

Thanks! Which stack do you use?

NoGo criteria NR1. No test/test frameworks included.

I know, I know!! I will include unit tests... some day... soon.

Does anything exist like this that uses a Java backend?

To be honest, I don't know. Probably no - but maybe just because Java is much less popular language than Node and Python?

Maybe try https://www.jhipster.tech/?

Looks interesting although it seems to me not SaaS-oriented.

Any plan to add job task management like RQ or Celery?

YES! As soon as anyone will ask for them :) You can add your request (with the use case) as an issue in the repo.

Update the license, now it's MIT.

Reply on news.ycombinator.com