I’m hoping someone can shed some light on a problem I’m facing.
I’ve invested 14 hours into this project non-stop with no sleep, but unfortunately, it STILL doesn’t work and my time feels wasted. I’m struggling to figure out how to make this function properly.
My goal is to navigate to an already open Instagram tab at the start of the automation by locating the favicon image in the browser tab, which worked.
Once there, it should go to the search bar over to the left, look up ‘Photo Manipulation,’ click on the first result, then click on the first image to bring it into full view.
It should then take a screenshot via a Mac, upload it to Chat GPT, create a comment based on the image, paste the comment, and post it before moving on.
Despite adding random delay triggers, the process gets stuck after the search step—it doesn’t click on anything. I even tried incorporating an “enter” command after it types in the search term, but nothing seems to help.
I’m really frustrated and on the verge of giving up on all of this.
Any advice would be greatly appreciated.
I have read and even transcribed every single tutorial that was made on this software and I still don’t even understand half of what any of these things can do.
I create a video for you! I think it will help you.
I created the scenario and also explained what was the changes I did differently then yours.
Here is the scenario, make sure to create image parameters as explained in the video.
Click Text “Instagram”
Click Image {InstagramSearchIcon}
Write “#photomanipulation”
Click Text “#photomanipulation” [1]
Click Text “Top Posts” -YMargin 50
Click Image {InstagramLikeIcon3} -Double true
SendKeys SNAPSHOT
Click Text “Message ChatGPT”
SendKeys PASTE
Write “Give me an instagram comment based on the image. Don’t add anything else other then the command so I can copy and paste”
SendKeys ENTER
Click Image {ChatGPTCopyIcon}
Click Text “Add a Comment…”
SendKeys PASTE
SendKeys ENTER
I already checked this out, but it wasn’t what I was looking for.
Basically, the flow goes to Instagram, opens the main feed, screenshots a post, gets a response from Chat GPT, then it likes and comments on the post accordingly, and then for some reason randomly leaves “dummytext” in the comments and keeps scrolling.
I’m really confused about how the scroll depth works. There’s no explanation of how it’s measured.
So, I have to pick a random number like “10” for instance.
But what does “10” mean? Is it 10 pixels? Does it scroll down in 20 pixel increments? I don’t understand how the scroll feature works after that tutorial. I did try to implement it though, it just confused me more.
Scroll is like a mouse wheel scroll. It changes from Pc to Pc.
dummytext is added to make sure that there is no more “Add Comment” on the screen because it can write text to an already commented post. This is to avoid such behavior.
Check it out. I created a series of automations and nested them all into a job, so now all I have to do is randomize this and duplicate the scenarios within this job until I reach 10 comments and 5 likes.
Then what I’m going to do is create a duplicate of this job 4 times and just randomize everything so that it feels like a natural human is doing it.
Then I’ll schedule each one to run at the top of the hour at specific times like maybe 12am all the way up to 4am.
I feel like I’ve mastered this thing finally. I tested a few and I’m just laughing because now my speed is going to feel like a fucking Jedi Warrior.
I’ll post back the results in the shared use cases.
Thank you for helping me spend the time and going through everything.
Update: I tested your code Zek but it needed some adjusting. Instagram hit me with a wildcard and it wouldn’t scroll to the next post if I used the keystroke “right arrow” after I commented on a post.
So learning what you taught about the X and Y axis, I was able to locate the smiley face icon that let’s you select emoji’s for your post and tell it to click -95 on the X axis, THEN hit the right arrow and this worked.
Instagram kept the blinking cursor in the comment and it was not allowing me to scroll to the next post. So I had to click a random area on the image to “deselect” it, and then it scrolled past.
From testing the full automation to success, it was actually quite funny to see the reactions from the creators. They were grateful for my comments and I gained a follower.
This was a small controlled group of just 5 auto comments. I can’t imagine what type of results I may get if I do 40 comments a day. I tested the “like” strategy you suggested and it worked perfectly, and I did it in a few other instances and it definitely helped.
Here is my code for commenting and moving on:
Delay 3000
SendKeys CMD+CONTROL+SHIFT VK_3
Click Image {ChatGPTfavicon}
Click Text “Message ChatGPT” [0]
Delay 1000
SendKeys PASTE
Write “Give me an instagram comment based on the attached image. Focus only on the biggest image shown here, then craft a witty yet positive art critique comment. Phrase it as if it’s an instagram comment from a young 35 year old graphic designer from Long Island, New York. Do not include hashtags, keep it short in 2 sentences, make it sound more natural and friendly.”
SendKeys ENTER
Delay 20000
Click Image {CopyGPT}
Click Image {InstagramFavicon}
Click Text “Add a comment…” [0]
SendKeys PASTE
SendKeys ENTER
Delay 1000
Click Image {SmileyFace} -XMargin -95
SendKeys RIGHT
Delay 1235
I used a separate scenario that launches Instagram and does the whole search functionality, that way I can quickly adjust and alter the flow so that it searches for other hashtags on the web.
My strategy will eventually be to do 10 comments in 10 different hashtags spread out over a period of time so I don’t get suspended from Instagram. I added some delays in order to trick Instagram’s algorithm on the timing to make it seem more natural and also I noticed some extra loading Instagram does.
Hi Zek,
I wonder if we can do the same with Facebook where the messages that we want to respond to or comment on are text based and not picture based?
I am exploring Uxer now so forgive me if the question is too basic.
Thank you in advance.
Educe
Hi @uxer3, Yes of course, and now it is much easier with AskUxer action which we added recently.
I didn’t try this but it should work or help you do it much faster. I will share similar use cases with a video soon!
You need to create image parameters ( {CommentIcon}, {Response} )
Openbrowser Chrome “facebook.com”
Scroll 2 -Untilvisible Image {CommentIcon}
Click Image {CommentIcon}
{Response} = AskUxer “Give me an instagram comment based on the image.” -UseScreenshot true
Write {Response}
Sendkeys Enter
Hi Zek,
Thank you for your quick reply. But in my scenario, that is Facebook and the posts on Facebook are mostly text based, not picture base. Do I need to change the script?
I am looking forward to your video!
Regards,
