I currently use n8n, but always great to see more alternatives in this space.
One thing that has surprised me is that no one has tried to standardize on the Zapier or IFTTT APIs / SDK implementation to allow for faster deployment of integrations. Seems like there could be a movement here that is akin to how many services use S3 compatible APIs.
We checked their abstraction, but it was not a good match with what we have in our minds. Apart from that, it also differs for architectural reasons because they don't have to make it self-hostable, which also affects how they build/run the integrations.
We noticed that too and thought scripts with simple main functions and their parameters could be the universal layer for integrations, so we (Windmill) built an open-source hub for it https://hub.windmill.dev
What do you believe your biggest challenges to adoption are? Also, Zapier’s value is in the long tail of integrations and their staff who keep those integrations working for their non technical user base when there are breaking changes (expected and otherwise). How will you solve for that? (Recently I thought ChatGPT might be a fun possible solution when Sentry gets the exception from an integration runner, and it could propose a code fix for the on call SRE or platform engineer, something to consider)
Best of luck, big market with a lot of dollars to go around.
Imho we need to separate the customer base here to think about the adoption. If anyone (people/company) would like to have automation but is also okay with using cloud solutions, it would be too difficult to convince them to use Automatisch rather than Zapier, Integromat, or any other. But if they don't have the option to use cloud solutions with the compliance policy of their company or just because of the sensitive user data, they need to check open source solutions. So those customers will be our main target. While we're doing that, we don't limit the open source version in terms of integrations but create another license to separate the feature set that will be only required by corporate companies.
We wrote one of the actions in our integrations by using ChatGPT. So action was easy to build, but still, we would like to utilize AI for those tasks.
Neat! Do you plan to actively sell enterprise licenses to mentioned industries, like healthcare and finance? Is that how you'll make money?
Random idea: Maybe sell integration building as well? Instead of just being able to request a integration. For example, if I would want to run this as a marketer and self host (for whatever reason), I would need integrations such as Facebook Lead Ads, TikTok Lead Ads, Gmail, etcetera. But instead of requesting them being built, I can imagine some people would rather pay $1000 to have them built _guaranteed_ within let's say 1 month. Because if there's some integration missing that is essential for my business, requesting and waiting 12 months until it's _maybe_ build is probably not an option. At least mentioning this service on the landing page and docs will cost 5 minutes of your time and maybe get some business. :)
Yes, we plan to sell enterprise licenses, but we also have other services, like prioritizing the integrations, as you suggested. The integrations prioritized will be available to all other users, but the companies don't have to wait until it's becoming most-requested integrations in the community.
This looks very cool and I'm excited to try it out later today. In the past I've tried Huginn but I have always found Ruby to be particularly unapproachable and the first integration I tried was broken (the weather integration) and despite submitting a PR to fix it, the PR has languished for years.
Here's some questions:
- How does this differ from Huginn?
- How easy is it to install integrations from outside the main source tree? In other words, can I create an integration and publish it on my own Github profile and have others use it with ease?
- Is Automatisch a purely OSS project, or are you planning to build a business around it? What type of business model?
- Is Automatisch primarily focused at home users or business use cases? The integrations listed in the docs seem primarily home-focused.
From gleaning at the repo's license, they use AGPL. So that seems tailored towards building a business around it. I find that quite exciting, because I have been thinking about which license to use for a project of mine as well, so that a) it is open-source, and b) I can build a business on top of it. AGPL seems to fit the bill perfectly, as long as you are careful about the license of contributions.
Is technically open source the best kind of open source? It is open source according to the OSI, but many including myself are interested in open source for specific reasons that the AGPL doesn't satisfy.
- It may not differ in the sense of integrations and general functionality. However, the most significant difference is that we aim not only for developers but also general users who may not have programming knowledge.
- At the moment, it's only possible if we merge the integration into the source code, but we're thinking about possible ways to have external integrations and also private integrations.
- We would like to build a business around it without limiting the open-source version in terms of integrations and core functionality. It's still in the plan, but I can say that only the enterprise feature set will require payment.
- We focused on mostly business use cases, but we also have home users, and we're happy to support them.
I use n8n too, but the biggest problem for me (as a developer) is, that it cost too much time to define complex rules and test them. An easier define and better test output procedure could be a great advantage over n8n.
Don't concentrate too much on "non" developers, cause these people normaly not use such tools, no matter how easy they are. That is why i hope you add better tooling for developer, to make their live easier.
This looks great. I really want a way to plug in to (in my case) Xero, and do some automation. I'm perfectly happy writing code but it seems painfully complicated to do all the oauth authentication stuff and hosting an interface and whatnot. All I want to do is trigger an event on something happening. Is adding oauth services relatively easy?
Great stuff! FWIW I've been looking for a "some code" platform, that removes the overhead and plumbing requirements for automation, but still allows arbitrary code to run in the automations. Basically, I can code, but there are loads of things about web development I have no interest in or want to think about. What would be amazing to me would be a way in which APIs can be abstracted away and a low code API for various languages is provided. It could just be JSON or something, but without any crappy hosting or oauth concerns.
Totally agree, we were looking for a similar “some code” approach. Automatisch and other no-code automation products (Zapier, n8n, etc) are awesome for non-technical users to automate simple tasks. We started building Workflows (http://retool.com/workflows) because we were looking for something faster than writing cron jobs/spinning up more infra. Products like Zapier, often times weren’t customizable enough without the ability to really write code or secure enough to connect to something like our production databases. Would love any feedback!
Thanks! We might build a code step or something similar to write a basic JS function to have a custom action/trigger but not sure when it will happen because our main focus is to allow our users to build automation without writing any code at the moment.
If it's obvious that the product won't get approved by risk management due do missing essential policies, I won't even take a glance at the actual product.
Automatisch would be interesting for me exactly due to:
> It allows you to store your data on your own servers, which is essential for businesses that handle sensitive user information and cannot risk sharing it with external cloud services. This is especially relevant for industries such as healthcare and finance, as well as for European companies that must comply with the General Data Protection Regulation (GDPR).
But automatisch.io doesn't include any data protection policies despite advertising a paid enterprise offering on their website and asking for data in a typeform. Also, there is only a copyright symbol without any link to a legal entity.
If it was just a non-profit open source package, I'd still look into it. But as it is, the website screams „don't touch this with a ten foot pole“ from a perspective of someone who is supposed to be the target group of your product ;)
They work at Automat, or worked there at some point. Have you any information that there is more relation than that? And at the moment they seem to not making any money with the project, so there is a limit to what law demands from them at the moment.
I think you can start with what you would like to automate. It might be difficult for individuals to come up with an idea, but companies tend to have manual/recurring tasks which they spend time and money. So those tasks are the potential pain points that needs to be automated.