Is the product polished? We need your inputs

I think ActivePieces has improved a lot in the last year. I use it for personal-use small automations.

In my experience my automations have broken many times because of updates to pieces (for which a could not find any documentation) and rendered my automations useless. I had to, after trial and error, fix them or adapt them (when my use case was no longer supported) in order to use them again.

Some examples:

I have an automation that automatically posts to a telegram channel. Sometime ago that did not work anymore and I realized that there were new options in the telegram piece, such as format (markdown or html). After researching and reading the telegram bot api documentation, I had to include a js piece to pre-process the text and make it telegram-compatible.

Last month, I. The same automation, the LinkedIn piece stoped working. I also researched and tried to make it work, this time without success. As this is not a “business” automation I just activated the new option to ignore the piece error on the automation and continue.

What I mean with this is that for professional or business cases, where a company or person hope to have a reliable automation system, this kind of behavior is far from polished.

I also use Zapier for the company I work for and they email me whenever a piece I’m using (a zap) has a breaking change or will no longer be supported. In the cases where a piece will no longer work as expected, they propose alternatives. It does not always work but it’s by far a better approach that leaving me to research on my own.

In the example a gave above I did look here for other cases, I opened an issue (without response) and I researched a lot.

If some pieces are not that important for you as a company (understandable from a business view) then the piece documentation should state that.

I also think that the question of whether the product is polished enough gets a different response from me depending on the use case.

For a business case, after my experience the last few months, I would like but can’t recommend it to my employer.

For e personal case I think it does the job, but by error it required a bit of a technical background to solve (ignoring the fact that is open source and a dev could make a pr)

For a professional case, it’s similar to the business case. Knowing my experience so far I will be testing this for some freelance projects, but under the premise that some thing might break.

I hope this helps :slight_smile:

3 Likes

@daco Thanks for your feedback!

Did these flows break without you touching them? If this is the case, I’d like to collect more info/evidence to help our team avoid this in the future. This MUST NOT be the case, as we want to make sure your automations are all reliable all the time. And sorry for that experience.

Is the reason why you won’t recommend Activepieces to your employer solely that the flows break or is it something else? You answer will help us a lot

That makes sense, and I see the distinction between “small easy fixes” and “bigger systematic issues”, but that’s not what you asked for in the start of the topic. You asked what was forming the perception of unpolishedness, as far as I can tell that’s exactly what I described.

@daco’s (much better worded) comment corroborates the (potential) issues I encountered during my testing so far. Activepieces just isn’t reliable enough to use in a mission-critical environment yet, especially with non-technical people involved. Given the open-source and young nature of the product that’s not a terrible thing though, as long as we’re all working towards improving the situation, which absolutely seems to be the case! I praise the team’s transparency in this matter, as evidenced by the fact that they started this public community discussion based on an off-hand comment I made in the Discord :stuck_out_tongue:

However, I do recognize that my earlier comment could easily be misinterpreted. As mentioned before I’m incredibly enthusiastic about using Activepieces, so I’ve edited my comment to more accurately reflect that. My only intention is to share my perspective, in the hope that we can collaborate on creating a better product!

It’d definitely be a good start, but is there a reason to not just track every-ish modification, similar to Google Workspace? (yes, see below)
Undo-stacks being lost after a refresh (e.g. caused by a network issue) is a fairly common and incredibly annoying issue I experience with other products :smiling_face_with_tear:

This’d also make figuring out the cause of issues a lot easier, analogous to git blame and bisect. If I mess up but only notice a couple hours/days later I’d love to still have that history.

Edit: I suppose this’d involve expanding the backend+database instead of just storing it client-side in memory or localstorage (which could survive a basic refresh). I still think this is an essential feature long-term, but a short-term basic version is already much better!

Temporarily disabling actions would be amazing, especially for things like emails and destructive actions.
The result of the last non-disabled (test) run could potentially be used as a “dummy response”.

The reason I asked for a specific modifiable dummy response (vs inheriting the last non-disabled test run,
if such a feature was available) is because sometimes you can’t send realistic looking data over through a connection (for whatever reason) and/or you never set up a connection in the first place, but would like certain data available (in a specific format) to test another step @HoldYourWaffle

Also, @daco well said. Reliability is of the utmost importance

1 Like

Hi @ashrafsam

Yes, the flows broke without me doing anything. I noted they broke because one (that published the last rss feed item to telegram and LinkedIn) stop doing so. I received an email that something went wrong the last time though. That was better than nothing.

And yes, that is one of the main reasons I can’t recommend this to my employer yet. There are some missing pieces (for example YouTube upload last time I checked), but the main reason is that we need something that does not break on its own. I understand that active pieces is still new and all this is part of the process. But the change to the community instead of help desk was both interesting (because of the community) and also I new reason for me to not use it in business, as I find it hard to get support. (Last year all I needed was to send an email and I got, within a few days, a response Now I posted an issue and it got closed for no activity).

I hope this helps you further.

3 Likes

Flows breaking by their own is definitely not something that we’d like to accept, we’ll discuss this internally and reflect on the incidents we’re aware of and see how we can further prevent this. Before your comments, I’ve been under the impression that this is not something that could possibly happen on Activepieces (by design).

Now, I need to go through it again with the team to see where we stand and how can tackle these reliability issues.

Regarding support, @daco we offer email support under our standard Enterprise plan and more hands on support under custom enterprise agreements.

Thank you everyone for working with us to take Activepieces to new heights, our team is capable of meeting your expectations and we’ll double down our efforts in doing so.

Related: Convert mp4 to binary format - #2 by S_M

1 Like

Hi @daco,

This really shouldn’t happen. Can you confirm if you have republished the flow and it started breaking, or was it working one day and then it stopped the next? Also, could you please tell me when was the last time you faced this issue?

By design, this shouldn’t occur, as we only auto upgrade in DRAFT and it’s supposed to be compatible. The published version should never change. From what I understood, the Markdown option wasn’t compatible and you are correct. I just took a look at the PR (feat: add format to Telegram by tanoggy · Pull Request #2598 · activepieces/activepieces · GitHub). It seems the default option was text in the Telegram API, and the new property default is Markdown. So, that makes sense why it breaks in your case, but it should only happen if you republished the flow.

I’d love to investigate the LinkedIn scenario and see how we can automate this review to avoid similar cases. You’ve given me some ideas.

Thank you,
Mo.

2 Likes

Thank you all for your replies, I would love if you give us inputs specifically for our Automation School by going to this post.

Another instance of instance of unhelpful error handling: Connecting Leadconnector (High Level) account "type error cannot read property clientkeys of null"

Is there perhaps a better place to track these?

Here are the things for me that are missing that will make the platform more polished:

I can only compare it to a more stable apps such as Zapier and Make.

  • Instability - there are times when I go back to a flow, and a piece would be missing some values/parameters, so I have to go back and re-enter some of the values. This happens from time to time. See screenshot below.
  • Performance issues - having to wait 10 seconds or more sometimes so I can re-test a step while it’s saving is a pain point.
  • Flow versions can be confusing and doesn’t work as expected. For instance, how does a new version gets created. My assumption is, when I make a change to the flow, but even if use a previous version as a draft, it doesn’t remove the most recent version.

A few enhancements that will increase productivity for me:

  • Disable a step - I have a workaround for this without deleting a step but it’s not ideal.
  • Re-usable step/piece - To be able to re-use a piece across multiple flows based on some settings or configuration would be nice.
  • Common used piece - It will be nice to have a separate tab for “favorites” or “commonly used step” would speed things up.
  • Improved error handling - If a step fails, perform certain actions such as a separate branch (think of try/catch).

As you can see in this screenshot, there’s no way that the last step would be empty as it already ran several times already today.

2 Likes

Agreed! I think all of your points should be considered :+1: @menacestudio

1 Like

@here We started making announcements about the points in this thread, here: We're polishing Activepieces — a rolling announcement

Thanks a lot for the GREAT feedback! :smiley:

1 Like

I’ve experienced this as well, seemingly every time I go back to my experiments on AP Cloud I have to reconfigure half of my flow, very strange.

This is a very important one for reliability. I need to be confident that my flows are not going to fail silently, screaming for help > confidently doing the wrong thing.

1 Like

Another instance of unhelpful error handling: Append Row to Worksheet error in Microsoft Excel 365

That’s what known issue is for, I moved it there

Found two more while testing today:

  1. It’d be nice if we could double click a step on the canvas to rename it.
  2. Maybe I’m missing something, but there seems to be no way to quickly switch between multiple runs. Opening a run replaces the navigation panel, but there’s no “Back” button.

Great work so far, can’t wait to see Activepieces grow further and get better and better!

I just noticed that the overall “crampedness” of the interface is reduced significantly by zooming out the UI to 80-90%. It’s not perfect, but now that I’ve noticed it I can’t unsee the fact that most elements just seem a little too big for the available screen real estate.

Here’s a comparison, though I’d recommend trying it yourself:

80% is a huge improvement for my combination of screen and eyes, though occasionally some pieces of text can become a little hard to read.

Something that need more attention it to improve online documentation for the input format of some value.
A good example if Google Sheet insert multi-row. I spent ~30minute to figure out the proper format for the input values. Proviving a very small sample of data inline would be helpful.

e.g.:

3 Likes