I have a problem, or a question, and then I 'google' the answer, so "web search" - and I must say that the addition of AI to Google searches has made it 10 times more powerful/quick/useful (this is an area where AI really shines) ...
I usually start in the AI response, then punch out to the cited resources or start scrolling through the results below, but it just depends on what I'm looking for or what kind of responses/results I get.
Most of the time I don’t go looking for new tools. They find me. A post here on DEV, someone mentioning something at a meetup, or a colleague at work saying “have you tried this?” That’s where most of my discoveries start.
Then I search for it the old-school way first, because I want the official docs and the GitHub repo, not a summary. I check when it was last updated, how many open issues there are, and whether people are actually using it.
After that I use AI, but mostly to explain it to me like I’m five, or to compare it with the alternatives. I don’t trust it to pick the tool for me, because sometimes it confidently recommends things that are outdated.
And the final test is always building something small with it. If I can’t get a tiny project working in an evening, I usually move on
Not search-first or AI-first for me: complaint first.
I don't go looking for tools. I go looking for the same complaint twice. Someone says "we keep losing track of X" in a comment, then says it again a month later with the same shape. The second time is the only real signal that a tool will get used. Then I read the docs and the open issues myself, and I use AI to widen the net, not to decide.
A feed of new libraries is a list of things nobody needs yet. The repeat complaint is the only part that pays.
Mostly a hybrid for me. I start with a normal Google search to see what's out there and who the competitors are, then I use an AI assistant to compare options and summarize the trade-offs.
I still go to the product's own site or GitHub before trusting anything, because AI summaries can be out of date or get details wrong. I also check recent discussions on places like Reddit, Indie Hackers and dev.to to see how real users feel.
So: search to find, AI to compare, and the source itself to verify.
We're a place where coders share, stay up-to-date and grow their careers.
I think one of the most common patterns is:
I have a problem, or a question, and then I 'google' the answer, so "web search" - and I must say that the addition of AI to Google searches has made it 10 times more powerful/quick/useful (this is an area where AI really shines) ...
I usually start in the AI response, then punch out to the cited resources or start scrolling through the results below, but it just depends on what I'm looking for or what kind of responses/results I get.
Hybrid for me, but in a specific order!
Most of the time I don’t go looking for new tools. They find me. A post here on DEV, someone mentioning something at a meetup, or a colleague at work saying “have you tried this?” That’s where most of my discoveries start.
Then I search for it the old-school way first, because I want the official docs and the GitHub repo, not a summary. I check when it was last updated, how many open issues there are, and whether people are actually using it.
After that I use AI, but mostly to explain it to me like I’m five, or to compare it with the alternatives. I don’t trust it to pick the tool for me, because sometimes it confidently recommends things that are outdated.
And the final test is always building something small with it. If I can’t get a tiny project working in an evening, I usually move on
Not search-first or AI-first for me: complaint first.
I don't go looking for tools. I go looking for the same complaint twice. Someone says "we keep losing track of X" in a comment, then says it again a month later with the same shape. The second time is the only real signal that a tool will get used. Then I read the docs and the open issues myself, and I use AI to widen the net, not to decide.
A feed of new libraries is a list of things nobody needs yet. The repeat complaint is the only part that pays.
So I take it you aren't into "listicles"? 😅
Listicles are for people who haven't been burned yet. Once you're burned, you don't want ten options. You want the name of the thing that did it.
i only discover stuff when i'm making a project. i let the ai dig up the library and teach me and then i just use it in my program
Can you talk more about this? Do you ask it for recommendations? Or do you select the first one it recommends?
i let it dig up whatever based on a few requirements and i try them one by one. first thing i like i use
I have subscribed to couple of newsletters that give me a steady stream of new shinny things i can look at... linkedin is another favorite
Google and here on dev through posts!
Are you scrolling through the Google results or just stopping at the AI summary?
AI summary is getting better at suggesting good sources!
So, both!
Mostly a hybrid for me. I start with a normal Google search to see what's out there and who the competitors are, then I use an AI assistant to compare options and summarize the trade-offs.
I still go to the product's own site or GitHub before trusting anything, because AI summaries can be out of date or get details wrong. I also check recent discussions on places like Reddit, Indie Hackers and dev.to to see how real users feel.
So: search to find, AI to compare, and the source itself to verify.