Now, I can still see work out there.
It's just more of a question around whose going to survive, and what will their roles look like?
@marcin.com.au
Coding since '98 from JS to React. I build projects & guide new devs. Follow for a journey of creation! π¨βπ»π οΈ Projects: ποΈ Youtube: https://www.youtube.com/@codewithmarcin β Building: http://teamegastore.com/
Now, I can still see work out there.
It's just more of a question around whose going to survive, and what will their roles look like?
Lets say all devs out there start really leveraging AI tools and they start deliverying work at a crazy pace.
What does the industry look like?
Is everyone now deliverying crazy amounts of work, or have teams been downsized and restructured.
I suspect this is the reason for hiring freezes.
Putting my business stakehoder hat on. Here's where thing get spicy.
Lets say the teams pumping out work.
I now have to worry about keeping the team fed with work otherwise they need to keep themselves busy.
And if they become idle or I can't keep them fed with work. Do I need to adjust the team?
I'll give you an example.
Lets say we're building feature with some tricky scenarios.
Without AI tools estimate is 3 days.
With AI tools (ie. screenshots and prompts), even when the generated result isn't perfect, its now less than a day to complete.
More work, faster. Awesome right?
That was my original thinking and then when I saw the posts about Saleforce and Zuck talking about hiring freezes and my mind flipped.
For example, I'm involved in large projects that don't factor in AI tools.
When I do factor in AI tool usage, even when not perfect. The time required is slashed.
Tried Googles notebooklm.google.com yet?
Thought I'd try feeding in some docs into it about "Effectice LLM Agents".
Two really cool Notebooklm features:
1. It allows you two generate a mini podcast of two AI hosts
2. The Interactive Mode lets you ask questions while the agents talk π€―
Are you a developer that converts designs into UIs?
If so, do this especially if you haven't before.
πTake a design and screenshot a feature
πFeed that screenshot into bolt.new and ask it to generate the feature
How close did it get?
We all have some serious soul searching to do in '25.
2025 hey? π€―
Slowly getting back into the swing of things. Slowly getting out of holiday mode.
Been tinkering with a few different workflows for my TeaMegaList directory.
Trying a new meta prompt for generating images and man, do the images look good.
Is there a AA for domain name buyers?
Just bought braindumply.com... π§
Will do!
Iβve got some ideas I want to try out tonight. π
Looks like @stackblitz.com has now released their official @supabase.com integration.
Might have to give it a test run tomorrow.
Conversation that I had tonight.
Person: The rules have changed.
Me: Ok, what are the new rules?
Person: We can't tell you
Me: Who can?
Person: This other person
Me: Cool, when can they tell me about the rule changes?
Person: We don't know. We're working on it...
π€¦ββοΈππ
Couple things Ayush.
All have their pros/cons. Serve different purposes.
Bolt/Loveable - In browser and great to get a fully-working MVP off the ground
Cursor/Windsurf - Code editors with AI for when you need to get your hands dirty coding
If you have an existing project. Go Cursor/Windsurf.
Gemini 2.0 is out, and there's a ton of interesting stuff about it. From my testing it looks like Gemini 2.0 Flash may be the best currently available multi-modal model - I upgraded my LLM plugin to support that here: github.com/simonw/llm-g...
Gemini 2.0 announcement: blog.google/technology/g...
Overall, pretty impressed with Gemini 2.0.
The it returns really deep results with Grounding turned on.
Can only imagine the type of projects people will come up with.
Really exciting stuff.
Good question!
Did comparison between Gemini 1.5/2.0 Flash with Grounding.
Asked for Gemini to give me a comprehensive list of public pools in/around the area.
Thoughts:
- 2.0 returns more results
- 2.0 returns a lot faster
- 1.5 results appear a lot more restrained
Check the screenshot out.
If you want to try Google Gemini 2.0 Flash out yourself.
Go to aistudio.google.com and you have a look at the screenshot to see where you can turn on Ground (aka Google Search).
Googles Gemini 2.0 with Grounding just blew my mind.
Been collecting the data for a number of directory sites I'm building and its taking time.
As a test, I just used Gemini 2.0 to do a search for all the coastal suburbs in NSW, and in seconds, I've got the data.
Imagine the possibilities... π€―
Thanks for the support @kulkarniankita.com. π
Googles released Gemini 2.0 Flash Experimental.
Use it for free at aistudio.google.com.
Key points:
- Free to use within limits
- Beating Claude (SWE-Bench)
- Large token count (Upto 1 mill)
- Supports realtime streaming (Voice, video, screen)
- Example projects - aistudio.google.com/starter-apps
Thanks Kevin. Appreciate it. π
Uploaded my latest Youtube video. π₯
In this one, "I Build a Directory Site using "Free Tiers" in v0, Bolt, Loveable and Replit".
Aim was to discover which platform gives you the most bang for buck!
Growing the channel, so a like, comment and subscribe would be amazing. π
youtu.be/Xd2VtN6SCKw
Congrats! π
Just a quick PSA.
Not all Australian wildlife wants to kill you.
Some are like this little nutter.
What are you building? :)
If your ever wondering what the different is between X and Bsky.
Its support from fine folks like this. π
Likes might not seem like much but hey, its a tiny push in the right direction.
That is until you press the db button too early and end up running out of tokens. haha
In all seriousness, they all have their pros/cons.
v0.dev - Minimalistic (Shadcn, etc)
Bolt.new - Great support for frameworks
loveable.dev - Fills in missing details (eg. design, toasts, animations, etc)
Going live for awhile.
Doing a Free Tier battle between v0, Bolt, Loveable and Replit.
Have any questions? Hope in and say hi!
twitch.tv/codewithmarcin
#buildinpublic
Ahh nice!
Nice lens.
Doing some landscape shots?