I read it happily waiting for the aha moment until the very end, where I realized that it was just an adpost for a new ifttt/transform tool with a glaze of "low code is the future" (All that for just $30k/year)
This was a pretty interesting adicle until the sales started in earnest. Tines looks mildly interesting, but it's wildly expensive and the community edition isn't available on-premises, so it's a hard pass from me. Plus, I'm immediately skeptical of any company that crows about anything Gartner-related.
At least for $30,000 per year you know they are also taking you for a ride, not just locking you in. I’d personally rather hire an assistant than pay $3,000 per year to have my lights turn off when I’m on a call.
Can you even imagine putting a price of $3000 per automation on a glorified IFTT? It doesn’t matter how much better it is.
But is there any reason you’d need this as opposed to any of the other options that cost less than 1/10th the price? My point isn’t that these apps aren’t awesome and worthwhile to use to automate things. Of course they are. But this price seems absurdly out of line with the market and value.
Yes. If you’re a developer that’s done a fair amount of automation, Python and the like are easier for batch or timed tasks. Having said that, I still use n8n for running event based source stuff which might still feed into a script. I also use it when I want a GUI instead of working out the “best” library and no HTTP API exists that I export by relocating requests from my browser (I’m looking at you all those devs moving to WS).
I just wish its main configuration/scripting language wasn’t JavaScript. If it was agnostic and I could drop-down to something else that would be great.
The Docker and packaged Apps they provide are excellent starting points.
I've used it a fair amount. Some cool features for sure. I love how you can copy and paste workflows from their docs / forum.
For some reason I can never fully grok how data is moved around from node to node. (Or at least remember when I come back after many months)
(e.g.: items[0].json.data)
The problem I have with it, is just not enough integrations. That's really the secrete sauce behind zapier. They have _so many_ integrations with a million different services.
From the outside looking in Tines looks to be a less feature rich and much more expensive version of Zapier (almost like a VC had the idea of building a Zapier clone and slapping a ridiculous price on it along with some "security" verbiage to spur the interest of larger businesses who see a $99/mo product and instantly believe it's not enterprise ready).
I do not see a way to automate most of my side projects given the free plan limitations and for an individual the pricing is a non-starter.
I know the founders of Tines, so I can confidently say that it wasn't dreamed up by a VC. The founders both come from an infosec background. They basically left their jobs to build the product they wished existed.
They were quite good at security operations. That probably made it easier to gain initial traction because they understand their target niche's problems.
I won't upvote this article because it's content marketing spam, but I respect the hustle and hope that it works for them!
I once had to implement SOAR in a “no code” automation solution (it was not Tines), and it was terrible. There were a lot of connectors and transformers, but almost all of them had some weird quirks that made them hard to use effectively. Coming from an “all code” background I was pulling my hair out trying to troubleshoot everything. The project was sufficiently onerous that I actually left my job over it.
I used to consult in the document capture space (OCR, classification, data extraction) and one company I worked for wanted to sell UiPath. I showed them how I could do all of that in code (even built a wrapper around AutoHotkey for Windows UI interaction) with way less time, effort, and hair pulling. They said it wouldn't sell because businesses think they can have someone implement a low code solution and they'll take it over eventually, and a significant portion of their revenue is from licenses.
I luckily no longer consult and now am an automation engineer for a company and I get to build things the way I want. So far a lot of Go and a little bit of Node and things are great. I'll take that any day over "We sell X, so you use X for everything."
That said, for home automation I do use Node Red and absolutely love it. The integration with Home Assistant is top notch and prototypes and tweaks take no time at all. It's much more like functional programming than a lot of low code tools and the function node allows for whatever JS you want. So if I need to do something complicated that would take a ton of individual nodes and spaghetti to work natively, I can just drop in some code and move on.
I had to do this too. I eventually learned I could write python plug-ins and started writing the workflows that way instead. But then it exposed to me how actually unwieldy the api was and I was extremely unhappy with the work.
I would automated everything myself, without the framework, but the platform was in part there to help those who lacked the technical skills to do everything from a terminal etc. I hear they’re still fighting with it even months after I left the job.
Co-founder of Tines here. This is exactly why we built Tines - I spent years working in security, and tried nearly a dozen different SOAR/SOAR-like solutions and felt the all were too complicated and the "connector" model was way too hard - the integrations were all quirky and limited.
Tines is completely different - we don't rely on any prebuilt integrations - similar to postman, if you have a curl command then you can paste it into Tines, and the response is simply the json from the API you're hitting - makes it a million times easier to do exactly what you want. You can then use that response in the next action. If the API is terrible (it frequently is) we've a ton of templates to help get the relevant data, and a community with experience connecting to most tools.
Would love to talk about the challenges and see if anything we're doing (or could do) which would have made it easier? I hope you're in a better job now anyway!
These comments all present valid reasons why low-code is usually a nightmare. To me, the most egregious offender is vendor lock-in: https://en.wikipedia.org/wiki/Vendor_lock-in.
It's all fine and dandy if you found a low-code solution that works for you, as long as you're comfortable paying a premium to pigeon hole yourself with that solution for the unforeseeable future, with no guarantees that you won't get royally screwed by some combination of planned obsolescence, feature deprecation, or the success of the company whose low-code product you are using.
Vendor lock-in is not always a problem. Imagine that you have a situation where you can either throw one engineer at it to create a custom solution or buy an existing one that locks you into their service offering? What would you choose? It is not always clear-cut, but it is better to buy simply because employees are not permanent in most cases. After they leave, there will be zero support.
> My main issue with Node-RED is that I have to install, run and maintain it myself. There have been times when I've forgotten which Raspberry Pi it was running on and I've re-flashed the SD card!
Installing Node-RED is a low-effort task. Setting up backups of your flows is a tad more effort, but plenty of documentation around this. It also does IMAP, I believe. In an article bemoaning vendor abandonment syndrome, something free that runs on your own equipment, is a feature.
A post-it note would be a cheaper solution to that second problem.
By all means, learn to automate ditch-digging. That is certainly useful. If it's also your job, perhaps you'll notice there's just a higher-order ditch to be dug. If so, what's the next move? Automate the automation of the ditch-to-be-dug, or start writing a blog about first-order ditch-digging automation? Or shill for a company who furnishes ditch-digging tools?
In which case, I'd suggest learning how to write in a way that doesn't channel Dan Brown blog-shit (section header every two paragraphs vs. two-to-three pages a chapter). Or learn to write ad copy, or talk about using VAX systems to fill out the preamble to your chili recipe. Even failing that, I suppose you've reached your no-code solution out of ditch-digging.
Co-Founder of tines here, thanks for all of very valid feedback, there’s loads to think about! I’ll respond to a bunch of the comments individually but the one thing I’d say is try it - it’s way more powerful and flexible than any of the products listed here, although it’s marketed as no-code it’s built by engineers for engineers, with tonnes of enterprise features (Retry on status, connecting to external password vaults, global functions and resources, MTLS, forms, free SSO, curl converter etc.)
As mandatum says above “The UX of Tines beats the pants off ALL of their competitors.” And we can talk about price.
Can you expand on your approach to pricing? $30k/year to only automate 10 separate tasks would be a non-starter for me. As a personal tool, the community version w/ 3 free is a nice way to demo things but too few stories to bother trying to run personal projects & conveniences on it.
My interpretation of your pricing is that you're positioning it towards businesses that only have a small number automation targets, but those targets might have a high level of complexity & are repeated very often. At least that's the main market target that makes sense to me given the pricing model.
As a product manager in this space, this stood out to me too. If I had to speculate:
- “Low code” is the current shiny thing
- These products are not sold to engineers, they’re sold to management types
- Management will only consider bringing in low code tools, because they believe this will save money on devs in the long run
- But that’s never how things play out, and engineers get involved anyway
So by making a “low code tool for engineers”, it seems they’re acknowledging the reality of the low code space and making sure the person who ultimately has to maintain this has the tools to make them successful while doing so…while also acknowledging the reality of selling integration/automation tools in 2022.
I could be way off here, but this is the only way that positioning makes much sense.
I guess it's meant to be low-code but as engineers they understand you can't low-code your way out of everything so escape hatches or advanced settings are provided for when "low code" isn't enough.
I once worked with a software stack called ActiveBatch. I really wish there was something as easy to set up as ActiveBatch available for the consumer market. Even something like Apache Airflow isn't competitive in terms of ease of setup.
Activebatch is literally just install the scheduler on a Windows system, connect a SQL Server DB(use community edition) and then install the "execution agent" on whatever machine you want to run jobs on. In 10-15 mins you have a full automation environment.
From there, you have access to a decent GUI where you can drag, drop connect jobs together and mix and match different technologies, APIs, whatever because you can actually pass data and context from one step to another. They have a library of hundreds of "Job Steps" that range from simple stuff like copying files/FTP/SSH to things like calculating checksums to things like interacting with Twitter to send out tweets based on whatever condition you specify. There is just so much you can do and then they allow you to extend it with scripting/APIs etc.
Has anyone here used Activebatch? It is a hidden gem. Unfortunately it is an enterprise product so it is off limits to most people. Really wish someone made an open source alternative. :/
Thanks for the suggestion. I'll check it out. Looks like it has a simple installer for Windows which is great but lacks a DB backend to store all job data and to operate the scheduler.
Maybe Huggin can do some of what you are interested in? https://github.com/huginn/huginn It also sounds sort of similar to Node red though that is primarily based around iot and api processing and routing.
Here is a related story. I should mention Tines or OP has no relationship I know of with Salesforce, for background Mulesolft is a cloud focused integration platform owned by Salesforce, who's big feature is visual mapping, visual workflows, etc. (of course it is more complex than this, read up if you are interested):
I attended another meetup/training, Mulesoft was really pushing their web only visual workflow/mapping tool. And the instructor goes over the entire song and dance, and I said to him, that I was having real difficulty even understanding what type of problem this tool would be good for in a real-world situation. And he takes his glasses off and says to a room full of maybe 30 or 40 developers, that this tool they had been pushing so hard should NEVER be used in production, its only real use was for sale demos and maybe, maybe, doing some sort of POC which would need to be reproduced in 'code' at a later time. He used weasel words to say all of this of course, but his meaning was clear. This was 2018~2019, not the stone ages. So, I wasted that evening (there would be more) going to a sales enablement seminar billing itself a developer learning workshop.
I actually had to use this about a year ago to build a SOAR system at a Fortune 500 company. The Tines team was pretty responsive to questions and requests for help, and they even sent me a nice box of swag. The box arrived after I resigned from that job, however.
No matter how many niceties you put on flow-based programming, the tools I had to use (all serverless) were the wrong ones for the job, and management was unreceptive to feedback (they wouldn't even give me a github repository).
I think Tines has potential but it's going to have to invest more into mitigating its shortcomings, not just further improvements on what it already does well. For example, as of March 2021, it was still far too easy to misclick and break something in production with no confirmation. You could export your setup as json, but IMO there was too much friction/manual work with that approach for it to be useful. To be fair, the company was pretty early stage at the time, maybe it's improved since then.
I don't blame Tines for what was the worst job I ever had (which I still have nightmares about to this day), but maybe someone else can be avoided the same fate with this feedback.
Thanks for the feedback! We've implemented a ton of features since then, most notably story history & versioning, annotations for documenntation, better alerting on changes (e.g. no event in 24 hours etc.), but recognise there's still a long way to go. I'd love to hear some more feedback if you want to DM me, it sounds like the project was a bit of a horror show!
I feel that all workflow tools are simply lacking in terms of capabilities. Sure, the base-cases are covered but when it comes to something remotely useful it is best to be implemented with real code. That being said, I think there is certainly a need for some low-code tools where your requirement is simply basic automation and service plumbing, as long as it is not mission critical, i.e. take value from system A and put in system B.
For example, do you really want to implement the entire Slack interactive callback workflow for displaying a simple dialog to a user? I guess not. I workflow will do a much better job at that sort of problems.
That being all said, I don't think tines does a particularly good job at this because it simply chains a number of HTTP requests. Some of these APIs require special care and HTTP simply does not cut it.
I have a tab delimited file that I bring into Excel. I need to review it on it a regular basis for cybersecurity reasons.
I load the file into Excel and then use the Excel Filters feature to get pulldown menus for each column. I then review the columns and filter for ports, user-agents, IPs, etc. Pretty typical stuff.
Anyone know how to automate the above that would basically just spit out files based on how I’m manually doing the filters? Then I can just open a file that is grouped by user-agents or open a file grouped by ports.
Starting at $30k, but that is the lowest paid tier!
I guess the thought is you can fire your developers and replace them with this. Ok I’m joking a little, but the sales pitch is probably that if you can hire one less developer, then it is worth it (I am skeptical, myself).
They’re not competing against other SaaS, they are competing against someone’s salary.
I have had a lot of fun and success with Keyboard Maestro for Mac and Shortcuts for iOS. For my work, I try to automate Terminal commands. I also dabbled a little with bash scripts. I love that Shortcuts lets me think programatically and flex those muscles for fun tasks.
Here's some stuff I have set up:
Mac:
- I use several displays, and don't usually use the built-in monitor. I have an automation that dims it until it turns off.
- I have recurring KM script that deletes Yarn and NPM caches. They get huuuuge.
- I have a bash script that shows me the temperature of my Mac. I might be able to add an alert if the temp goes above x.
iOS:
- I hooked up my Pavlok shock bracelet to Reminders. I get a percentage chance of getting shocked every 15mins if tasks aren't done. The more tasks done, the less likely it is.
- Shortcut to pull up Youtube videos for PiP without a Youtube Premium account. You can find that online. Game changer.
- Sci-Hub and Meta.org search shortcuts made available in the share menu. You can get my Sci-Hub shortcut here (a good starting point for other search shortcuts): https://observablehq.com/@iz/sci-hub-mirrors
- To help me debug shortcuts, I created a logger that writes to Data Jar or to a Note. Helps soo much. I haven't released it but you can get my Twitter at the link above if you're interested.
Node dev:
- Semantic commit template accessible through Dash's text-expander
- Next.js sucks but I made some scripts to automate that work. I had to use a Chrome-watcher that let's me refresh specific tabs from the command line.
There's many more. I wanted to share mine to maybe give you an idea of what tools to check out and what to automate to approximate your robotic luxury communist future.
There are a number of other (generally much cheaper) visual data transformation tools competing with Alteryx. For example: Easy Data Transform, Easy Morph and Knime.
This article is a lengthy preamble to an advertisement for a SaaS product, but I'm thankful you at least get to a conclusion before the evangelism begins: there's a basic pattern here, i.e. "get data from X, transform it, and put it in Y", and the tools, systems, and platforms listed are all solutions to the same basic problem.
I'm sure there's a use case and a market for this product, but I'd wager that I'm not alone among the HN audience in thinking that I'd rather have a couple shell scripts in a Docker container on my own hardware or whatever than spend a bunch of money irreversibly trapping my automations in some proprietary cloud-hosted thing where they can be held ransom come the renewal date.
This seems to explain a skepticism that I have, with regards to new tools, new languages, and especially "a new [x] that will solve all your problems."
And that basic pattern is usually called Extract, Transform and Load (ETL) and there are numerous implementations. Some respectable Open Source ones too.
I think there's space for an open source library that can help with what you described by using just python and YAML. We originally created https://github.com/typhoon-data-org/typhoon-orchestrator to orchestrate ETL workflows, which would be a superset of the use cases you described. Our next goal is to allow deployment to AWS lambda which can be a good compromise between getting locked in with SAAS and hosting your own infrastructure.
Also check out Zappa's scheduled tasks that have a similar goal and inspired our library. We used it initially as a backbone and ran into a series of issues that forced us to write our own version of it, but depending on your goals it could be enough.
