I’m a writer, not a programmer, so until recently a lot of the hype around ChatGPT’s abitilies as a coding tool went over my head.
But then I realized generative AI’s programming powers can be helpful for more than just coders. It can also help anyone else dabble in code to get things done.
In my case, that means creating new browser bookmarklets. These are special kinds of bookmarks that use JavaScript to modify or act on web content, and they’ve always been an underrated web browsing superpower. For years, I’ve used bookmarklets to speed up web videos, remove page clutter, and quickly search my favorite sites, but I’ve always been limited to whatever example code I can find online.
With AI tools like ChatGPT, I can finally make new bookmarklets myself, and the only limit is what I can think to do with them.
ChatGPT’s bookmarklet breakdown
Ironically, my “aha” moment with AI-generated bookmarklets arose while getting frustrated with another AI tool, Amazon’s Rufus shopping assistant. Last year, Amazon removed a feature that let you search through customer reviews and Q&As directly from its product pages, replacing it with the much slower Rufus chatbot.
That got me thinking about a faster way to search Amazon reviews directly. After noticing that Amazon has separate pages for products and customer reviews, each with the same product code in the address, I realized that a bookmarklet cloud allow for faster searching.
Here’s how I asked ChatGPT to make a bookmarklet that searches the customer reviews from an Amazon product page:
Here is a link to an Amazon product page, where the ASIN is B0DHV7LR12: https://www.amazon.com/Baseus-Charging-Certified-Magnetic-Retractable/dp/B0DHV7LR12
Here is a link to a page that searches through customer reviews for that product, where B0DHV7LR12 is still the ASIN, and “test” is the search term: https://www.amazon.com/product-reviews/B0DHV7LR12/ref=cm_cr_arp_d_viewopt_kywd?pageNumber=1&filterByKeyword=test
I want you to make a bookmarklet that, when clicked on an Amazon product page, opens a “Search Amazon reviews:” dialog box. The bookmarklet will then open the corresponding review page with the search term entered in the dialog box. If the bookmarklet is clicked outside of an Amazon product page, display an error message that says “You must be on an Amazon product page to use this bookmarklet.”
This link to my ChatGPT conversation includes both the bookmarklet in question and instructions for installing it. By adding it to your browser’s bookmarks bar, you can click a button from any Amazon page to quickly search its corresponding reviews.
More AI-generated bookmarklets
Feeling satisfied with my Amazon review search bookmarket, I decided to try making some others.
If you want to use any of these yourself, click on the links to each ChatGPT conversation, copy the JavaScript at the bottom of the chat, then create a new bookmark in your browser and paste the JavaScript into the address field:
Video Speed: YouTube’s speed controls take too many clicks to access, and I don’t like the default speed increments. I had ChatGPT make a speed-control bookmarklet to my exact specifications, and it works on pretty much any video site, not just YouTube.
Hide Stickies: Removes annoying web page elements that follow you around when you scroll, such as menu bars and floating video players.
Link Card: I use Obsidian for notetaking, and this bookmarklet converts web links into neatly formatted cards that I can paste into my notes.
Link and Excerpt: This helps speed up link sharing on social media. If text is highlighted on a page, clicking the bookmarklet wraps the text in quotes and copies it to the clipboard with the URL underneath. If no text is selected, it just copies the address instead.
More Links: Sends the current page to Perplexity with a request for more links to stories that cover the same topic.
Clean Link: Copies a link to the current address without common tracking parameters and other junk.
Google Maps Search: When clicked, this asks for your destination and starting address, then looks up directions. If you leave the starting address field blank, it just looks up the location instead.
I don’t expect you to use all of these yourself, but hopefully they’ll get you thinking about the kinds of things bookmarklets can accomplish, and how you might use AI to build your own.
One important note: ChatGPT sometimes inserts comment lines (denoted with a // double slash) to explain, but these can prevent the bookmarklets from working properly. Either remove them yourself or instruct the model not to include them.
Why this works
Generative AI is handy for making bookmarklets for a few reasons:
The stakes are low: While AI-generated code is causing all sorts of problems for businesses, here you’re just generating JavaScript to automate and improve your own web browsing. You’re not at risk of breaking anything critical.
The results are immediate: No extensive testing is necessary to see how your AI-generated bookmarklets perform. Either they work or they don’t.
They’re easy to modify: If you want to change some element of your newly created bookmarklets, you can just ask using natural language.
You might learn something: If you aspire to learn a little JavaScript yourself, bookmarklets are a simple application with immediate practical benefits. You can look at the code that comes out, compare it with other examples, and ask questions to understand how things work.
In general, bookmarklets work well whenever you want to perform an action on the current URL, modify web page content, or open a specific site’s search page with keywords pre-applied. If you’re not sure where to start, you can always ask your AI chatbot for ideas.
