What features should we build next?

Tell us what we should focus next on. Usability improvements? Are there features that you are missing?

Note:

  • This board is not an official roadmap. The primary space for feature plans, ideas, and the (often heated đŸ”Ĩ) debates about them is our #ideas-feature-suggestions channel on Discord.

  • This board serves as an additional space for those who aren’t part of our Discord community to share suggestions and feedback.

  • Once a suggestion is implemented, it is deleted from this board. We don’t maintain a “Done” or “Changelog” list here, but you can find a complete list of updates in the #releases channel on our Discord.

  • To stay updated on our roadmap and releases, join our Discord community: https://discord.gg/aSUhf75fSz.

Use prompts to create your taskbot (The future of automation!)

With recent releases like Jarvis ai (Google), Computer use (Claude), Co-pilot (Microsoft) - the market is heading towards AI agents that can do stuff for you just by prompting them. This is the future of automation. Zerowork shouldn’t be any different! If the platform stays in the same format where users have to compile each and every step manually, it will soon fall behind - as it’s much more complicated to pull off these things manually, rather than just explaining what you want to achieve in natural language. Please find a way to keep up with the latest technology, either via integrating with API (e.g. Claude computer use) or by adding the same mechanism inherently in Zerowork.

Jazzos 3 months ago

2

Detection prevention - VPN integration

It would be a great feature to be able to use a VPN to mimic a specific geographic locations for a bot. Even better if location can be set for a specific cookie setting (the cookie used to log in to a specific account on a particular platform) This way we can build bots that can hold say 10 different Facebook accounts (if we are so inclined) and then log in to each account from a preset geographical location every time. Example: 1 bot - 10 FB Accounts - Linked to your Proton VPN (free or paid). Accounts 1-4 log in from 4 different IP addresses in USA Accounts 5-7 log in from 3 different IP addresses in European countries Account 8-10 log in from2 different IP addresses in Asian countries. Now, we just added a new level of detection prevention. See my other ‘detection prevention’ feature suggestions if you too are keen to keep your bots safe 😉 Upvote to keep your bots running undetected 👍👍😉

Martin 4 months ago

Text typing - Simple and better human behavior

Detection prevention - When bots type text into a text field, it is like a monotone typewriter. No human types like this. To better mimic human behavior and hence avoid detection - A simple feature would be great to have. A simple feature that automatically takes the text it will be typing out. Adds some random words to the text in random places. Then as it types out the text it will type out the inserted words and then either instantly delete them using backspace or delete them from the text once the full text has been typed out. Example: Org text: “Hi James, I really like your post. Thanks for sharing” New text: “Hi James, I have {recently} like your post. {funky monkey} Thanks for sharing” As the bot types out the new text. It randomly chooses wether to delete and insert right away or if it will wait to delete until the full text has been typed. This will help our bots look like a regular indecisive human typing out the message. It would be great to have a toggle button to turn on or off a feature like this at the bot settings level.

Martin 4 months ago

Simple and better Human behavior mimicking

It would be great to have 2 global randomization settings for bots to make them behave more like humans. Improving bot detection prevention. This could easily be done by adding 2 types of ‘random wait time’ (set in milliseconds) settings for the bots. 1) ‘Random wait time between actions’ 2) ‘Random wait time between keystrokes’ Both set with a minimum and maximum number of milliseconds. By introducing these random millisecond based ‘wait times’ between every action and keystroke - no run of a bot would ever be the same in terms of how long the run will take. Making it even harder for the bots to be detected. Example: You have an outreach bot for LinkedIn - It looks for all the CTO’s it can find and sends them a personalized message. Right now every loop (send message) will be done in exactly the same manner and in very much the same time - so it will be very easy to spot for LinkedIn to build an ‘activity timer’ ⏱ī¸ feature that looks for ‘bot like behavior’ and then ultimately shut you down. But with these 2 global ‘micro delays’. You could set the action delay to 10-200 milliseconds and the keystroke delay to 5-50 milliseconds. For a bot with say 30 actions and an average message length of 250 characters. The bot would add between 1550-8500 for each message it sends and because it’s not just doing it as a single delay but as 280 individual micro delays - it will be much harder for any platform to build a detector that will be able to catch the bot based on its behavior. And let’s not pretend that with all the automation tools that are flooding on to the scene - That ALL the platforms we want to automate will implement countermeasures if they haven’t already done so. Upvote if you want your bots to operate without detection 👍👍👍

Martin 4 months ago

Option to "Save as a template" when selecting building blocks

When selecting a building block / multiple building blocks: Add the option to save the selection as a custom template (with its current configuration) then you can name the template to describe its function. These templates would be like pre-made flows, which you can always drag and drop into your taskbot (just like a building block) whenever you want to add specific functions. This would solve the need to either: create functions from scratch every time search in your taskbots for where you’ve previously configured a function, selecting the exact building blocks you’re looking for and then duplicate it to wherever you need it placed - too much cumbersome process (which is also prone to mistakes in case you’re not selecting the blocks properly)

Jazzos 5 months ago

1

conditional logic - check text on web element

I want to set conditional logic based on the text in a web element (e.g. if text = Xâ€Ļetc). Currently the only way to do this function is to save the web element to variables, and then start condition based on the variable - but that’s just too cumbersome. This is a function that should’ve been easily set in just one block, without having to use variables at all. My suggestion: add a “web element text” option to the dropdown selection in “start condition” block. Then, when “web element text” is selected, you can enter the element’s CSS or XPath selector in the field below. Then you can set the conditions in the next blocks as usual, according to the web element’s text.

Jazzos 5 months ago

1

Building block that writes all pending rows to Google Sheet

i would love to have a new building block that forces all accumulated/pending rows to be written out to the Google Sheet without having to wait for the next 100-row mark. i would mostly use it when i have a random delay built into the taskbot inside a loop (inserting it just before the delay). the longer the delay, the more likely it is that the taskbot can crash during that time, such as if the user closes the browser or shuts down their machine or whatnot. if that happens, i don’t want to lose the data that been generated so far. the reason i am not using a repetition limit on the loop and scheduling the taskbot to run every so often is that i do not want to keep starting a new run. in one case, it’s because each separate run will cause the site to show another captcha. in another case, it’s because i’m scrolling down a long page and i want to keep my position for the next run, and not have to scroll down and find it again. i also think that will reflect more human behavior. a side benefit (but definitely not a driving reason) is that this block can allow multiple taskbots to share the same google sheet at the same time, even when taskbot A writes to the google sheet and taskbot B needs to read that data before taskbot A finishes running.

zeroworkme 5 months ago

Send data table into Write JavaScript block

https://docs.zerowork.io/using-zerowork/using-building-blocks/write-javascript says “Custom looping through existing rows of a table within your code block in order to update data (as opposed to append) is currently not supported.” i want to set up a dynamic Start Repeat such that i can access two different rows of the table at the same time. right now, i can use variables to carry over the values from row X to be available when i am processing row X+1, but once i’m processing row X+1, i cannot access row X anymore. i’m guessing that if i can send an entire data table into the javascript block, then i will be able to jump around between different rows however i want.

zeroworkme 7 months ago

1