Find me on https://lovable.dev/@lazarjovanovic and learn Lovable on https://february.lovable.app/

Lazar Jovanovic

Professional Vibe Coder at Lovable | L5: Diamond Certified

Next stop - Lovabletown! ❤️ 

Dream it -> Ship it

Am I still crazy for saying no one will need to code 6 months from now? 
Am I still crazy for saying that PMs and designers do not need to learn how to code to build production grade software? 

Am I crazy? Or am I just trying to shake you from your roots? Make you do something about it? 

Not freak out by any means - but by all means take action and start doing something. 

It's entirely your choice and within your reach, please please please listen to Lenny Rachitsky and Boris Cherny dive deeper into this topic. 

Don't listen to me. Watch me as an example it is possible, and do something! 

https://lnkd.in/eK5jAdwq

Open video: Head of Claude Code: What happens after coding is solved | Boris Cherny on youtube.com

Head of Claude Code: What happens after coding is solved | Boris Cherny

Trying to come up with a good caption for this one.

This lesson is probably going to be very polarizing so I need your opinions on this in the comments. 

My POV:

The real leverage isn't in building more apps. 
It's in building fewer apps that can become many.

I used to think every project had to be unique. 
A new idea, a new market, a new everything. 
That is backwards.

Most beginners build apps. 
Builders build templates.

I don't mean a generic boilerplate. 

I mean a repeatable solution to a repeatable problem. 
Because most people in the same niche need the same things.

Take fitness creators. 
They all need workout delivery, progress tracking, and community features. When you build it, the pitch isn't "I have an idea for an app." 
The pitch is "I already built this. It just needs to look like you."

Or look at podcasters. 
They have audience and trust, but they lack products. 
You build a companion app once with searchable content and AI summaries. You use the same structure and simply swap the identity for the next creator.

The math changes everything:

Build one solid template in two weeks. 
Customize it for a client in two days. 
At $2,000 per branded app, five clients is $10,000 from that same initial week of work.

To find your own template, look for these four things:
1. A niche where people share the same needs.
2. A stable structure where only branding and content change.
3. A market where people are already spending money.
4. A clear way to reach that audience.

Your best work isn't just for one client. 
It is proof that you can deliver something premium, and that proof can be replicated.

Find a repeatable need, build once, and customize many times.

Catch Day 26 and the full lesson at https://lnkd.in/eWZShRpu and tell me if you agree about this being the easiest way to start monetizing your Lovable builds?

Orchestration is becoming more important than writing code.

1. AI agents are extremely eager. The moment you describe a feature, they want to start coding. That feels magical in the beginning because you move from idea to visible output in minutes. But once you touch anything slightly complex like payments, database migrations, authentication, multi-user state, that eagerness turns into risk.

2. Most people assume better prompts are the answer. In reality, structure is the answer. The difference between “fun prototype” and “real product” is not model quality. It’s whether you force the agent to slow down. Exploration first. Then planning. Then execution. Then review. Then documentation.

3. When I started, I let the agents build immediately. It worked until it didn’t. Bugs became harder to reason about. Changes started conflicting. Fixing one thing broke two others. That’s when I realized I didn’t need better AI. I needed architecture around it.

4. The biggest shift in the role of a software engineer and increasingly in product roles too is not writing code. It’s managing parallel agent threads, reviewing intelligently, and designing workflows that prevent chaos. The leverage is not in typing. It’s in orchestration.

5. Vibe coding without system design scales stress. Vibe coding with system design scales output. The tools are getting better every month. 

The people who win won’t be the ones who “use AI.” 
They’ll be the ones who build operating systems around it.

Professional Vibe Coding Engineer at Lovable | L5: Diamond Certified

Andreas Lehner

I think most people are still using AI at the surface level.

They use it to generate something faster. 
A document. 
Some code. 
A summary.
A deck. 

And that’s fine. 
It already feels magical.

But the real shift for me wasn’t speed. It was structure. I stopped thinking of AI as a tool that produces output and started treating it like a thinking partner.

Now I use it to pressure-test ideas before they become expensive.
1. To argue with me.
2. To ask me the uncomfortable questions.
3. To simulate edge cases I wouldn’t have thought about on my own.

The biggest unlock isn’t “write code for me.”
It’s shortening the feedback loop between idea and clarity.
That’s the part that changes how you work.

Vibe coding is just problem-solving under new constraints.

People hear the term and think shortcuts or laziness. 
In reality, it is the natural next step in the history of software development.

In Day 27 of 28 Days of Lovable, Marija Milosevic Lovable Ambassador and a former software engineer with 10 years of experience writing Java joins us to reframe how we look at AI-powered building.

If you zoom out, the pattern is logical:
- We moved from binary to functions.
- We moved from machine commands to readable languages like Python.
- We moved from manual coding to visual low-code builders.
- Now, we describe functionality in natural language and let the system generate the structure.

Every time we build on top of a new layer, we give up a bit of control in exchange for leverage. 

That has always been the deal. 
But faster execution does not mean less thinking. 
If anything, it means your thinking matters more. 

Thinking, judgment, and taste are what keep you calibrated when the tools move fast.

Over the last 27 days, the goal hasn't been to master a tool, but to master a loop:
1. Identify a problem.
2. Understand the constraints.
3. Design a system to solve it.
4. Execute as cheaply as possible.
5. Review honestly.
6. Decide deliberately.

When you look at an app like Uber, it feels like magic. 
But when you break it down into driver location, your location, distance, and a recalculation loop, it stops being mysterious. 

It becomes a system. 
And once you see the system, it is buildable.

Good builders don't wait for perfect tools. 
They see limitations early and design around them. 
That is what transfers. 
That skill doesn't expire when the tools change.

Catch Day 27 and the full lesson at https://lnkd.in/eWZShRpu

Community spotlight time! 

I asked -> you shared -> now it's my turn :) 

I cannot be more grateful for all of you that kept sending me Lovable app links for the past week. 

I hate singling anything out - but these are the 3 that popped for me. 

Keep sending stuff that you build and I'll do my best to feature you. 

https://meucv.online/ by Diego Ruzzinenti - simple (love that) application, the reason why I really loved it is because he took advantage of everything that AI development can offer. Check out the animation of how the app works on the page. That is a game changer. You no longer need to record videos of your demos or get people onto the waitlist. Just show them. 

https://lnkd.in/e8disuhJ by Erik Willemse - very cool for getting a bit of perspective in life, in case you ever asked yourself the question I've been asking myself a pretty decent chunk of my time on earth, which is, "Would my life be any different if I was born elsewhere?" 

https://lnkd.in/ezm3xJ88 Michael Tippett - Most people come to me, and when I ask them why they haven't used Lovable, they tell me they don't know what to build. I think this is one of the best apps to solve for the "cold start problem" that I've seen recently. 

Keep shipping, keep sharing, and I'll keep giving you the spotlight. 

Happy Friday!

Tomorrow, to end the lesson portion of 28 Days of Lovable - I have another surprise guest to announce! 

She's a Lovable Ambassador, a wearer of all hats one could put on their head and the best person I could think of explaining what I believe vibe coding is really about - problem/constraint solving! 

Stay tuned for an amazing 15 minute session from Marija Milosevic! I was on the edge of my seat for this one, and am convinced you'll be as well 🙂 

(and you'll hear a few amazing few quotes too)

Everyone talks about how AI makes you faster.
That’s true.
But something else happened that I didn’t expect.

It made me more exposed.

When you can ship in days instead of months, there’s nowhere to hide.

If the idea is weak, it shows immediately.
If your taste is average, it shows immediately.
If you don’t understand the user, it shows immediately.

Before, you could blame timelines.
Or dependencies.
Or engineering bandwidth.
Now the bottleneck is you.

AI removes friction.
And friction used to hide mediocrity.

That part is uncomfortable.
But it’s also the opportunity.
Just depends on where you stand. 

We're all producing faster. The question is what - magic, or garbage.

Most people use AI to generate a single image. Lucio Amorim uses it to build entire design systems!

Day 25 of 28 Days of Lovable, is special featuring him as the guest speaker for you. If anyone knows how to push Lovable to its creative limits, it's him. So I'm stepping aside and letting him take the stage.

Lucio's project features roughly 300 images, all maintained through a specific information flow where image generation is the very last step in a chain of decisions.

His core framework for consistency:

1. Use reference layers. When you stack a photo, a screenshot, and a logo style, the process stops being variable and starts behaving like a deterministic one.

2. Negative prompting. Telling the AI what not to do is just as important as the instructions. No text overlays, no busy backgrounds, no photorealistic style.

3. Character locking. By setting a base image reference and using a very low temperature setting, you can place the same character in different poses across different pages without the look drifting.

One of the most brilliant parts of this lesson is how he uses Lovable to emulate a panel of expert designers—a Minimalist, a Brutalist, and an Immersive designer. He has them openly argue with each other to justify edgy proposals for a brand. This generates genuinely distinct creative directions rather than minor tweaks.

By generating multiple distinct approaches and comparing them side by side, you figure out what truly works without endless manual iteration.

Catch Day 25 at https://lnkd.in/eWZShRpu

Lovable Ambassador 💜 Data-Driven GTM & Market Intelligence | 20+ Years Fortune 100

Most projects don't die because the idea is bad. 
They die because the structure collapses.

On Day 22 of 28 Days of Lovable, we are discussing the uncomfortable truth: sometimes a project isn't worth fixing. 

When fixes start breaking other fixes, you are no longer progressing. 
You are fighting entropy. 
And entropy always wins.

Starting over is not quitting. 
It is optimizing for leverage.

If you find yourself re-explaining the entire system to the AI just to make one change, you are past the point of efficient repair. 

The professional move is to extract what worked, identify what caused the breakdown, and move the good components into a fresh build.

You do not owe a project completion.
You owe yourself systems that can evolve.

Watch the full lesson on structural failure and how to restart intentionally in this video

I told you before - I am not great at design.  
Luckily, Lucio Amorim is one hell of a designer! 

And Day 25 (tomorrow) of 28 Days of Lovable is all about this topic. And how to use AI as your creativity amplifier! 

He is a Lovable Ambassador and a creator behind the app I shared this morning on all projects built by people in our ambassador community - a truly Lovable cause! 

I learned so so much from this video :) 

So happy that my little project will now feature more vibe coders - make sure you don't miss the episode tomorrow!

Lovable

For a long time, building something felt like training for a marathon.

You would prepare for weeks.
Write specs.
Align stakeholders.
Wait for engineering bandwidth.

By the time something shipped, so much time had passed that it almost felt detached from the original idea.
Now it feels more like going to the gym.
You have an idea in the morning.
By the evening, there’s a working prototype.
And that changes something fundamental.

When building is slow, you can hide inside planning.
You can say, “We’re still refining the strategy.”
You can live in Notion and feel productive.

When building is fast, that shield disappears.
If the idea is weak, you find out quickly.
If users don’t care, you know within days.
If the execution is sloppy, it’s visible immediately.

AI didn’t just make me faster.
It removed the space where I used to overthink instead of finish.

And finishing and shipping something real, however imperfect - is becoming the actual edge.
Not intelligence.
 Not credentials.
Just the discipline to turn thoughts into things.

Don't watch this video at all! 

Day 21 is about building, and this post is for my own consistency purposes - your only purpose for the day should be to create! 

And then visit https://lnkd.in/esnAfSXh to submit what you created as I plan to share it tomorrow. Stop reading! Go ship!

Finally on Android!! 

Shane Spencer let's get to 1M faaaast 😎 Bobby Siegfried can't be the only one!

I was literally thinking this morning - if we're all solving for software, how will the hardware keep up? 

Perhaps this is the answer!?

This result is just mind-numbing. When you can generate 1M toks/min (which corresponds to say 3-4k pages of a book, or a pretty large codebase) - what new AI applications become possible? :)

Ljubisa Bajic a fellow Serbian and former co-founder of Tenstorrent, is building a new AI ASIC at Talaas.

When we spoke last February, he told me this was exactly the direction he was heading in - and now a year later, he’s shipped it.

The ultimate chip specialization: burn the weights of the transformer directly into the chip. Why even deal with software (at least at that level of the abstraction)?

When you're providing the intelligence infrastracture at planetary scale (and given how much the LLM architecture has converged)  this can be a very viable economic model. You tape out a chip that can support 3-6 months of inference - and then you ditch it after that.

Keep in mind that these numbers are for an 8B heavily quantized model (so the quality is somewhat lower) - but then again this is just an "MVP". :) We'll soon be able to do this at a much larger scale.

Check out their blog post here: https://lnkd.in/gEXdXBKs

You can apply to try it out.

Most challenges reward outcomes. This one rewards process.

Today is Day 28 of 28 Days of Lovable, and we are closing the program by opening the 7 Day Buildathon.

Starting now, you have one full week to build and ship a project using Lovable. 

But I am not looking for the prettiest UI or the biggest scope. 
I am looking for clarity of thinking and constraint awareness.

I am judging five specific pillars of execution:

1. Deliberate Starting Point
Show why you chose your entry point. 
Did it come from a real problem, a remix, or a template? 
Day 1 through Day 5 were all about this.

2. Intentional Documentation
You must demonstrate context preservation. 
I want to see your Forward docs (what you planned) and your Backward docs (what you actually built and the decisions you made along the way).

3. Agent-Oriented Execution
Show how you designed for execution, not just prompting. 
This means structured task breakdowns and phased builds with verification steps.

4. Systems Over Memory
Demonstrate how you avoided relying on one-off prompts. 
Show the system you used to keep the project aligned as it grew in complexity.

5. Reflection
This is critical for builder maturity. 
What broke? What limitation did you hit? How did you work around it? 

A month ago, you might have hesitated to start or feared breaking things. 
If you have done the work, you are not that person anymore. 
You have practiced the loop: identify a problem, design a system, execute cheaply, review honestly, and decide deliberately.

That loop is yours now. 
It does not depend on any single tool. 
The thinking transfers.

Submissions close next Friday. Winners will be announced next Saturday.

Catch Day 28 and get the full Buildathon briefing at february.lovable.app

P.S. I am proud of you. And proud of myself for showing up every morning in February. It's now the best month of the year for me ❤️

taplio

Lazar Jovanovic's Linkedin Analytics

Lazar Jovanovic

Performance ScoreLazar Jovanovic's LinkedIn Performance

What is Lazar talking about?

Who is engaging with Lazar

Lazar Jovanovic's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Lazar Jovanovic's Linkedin Analytics

Lazar Jovanovic

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Lazar Jovanovic's LinkedIn Performance

What is Lazar talking about?

Who is engaging with Lazar

Lazar Jovanovic's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Lazar Jovanovic's Linkedin Analytics

Performance ScoreLazar Jovanovic's LinkedIn Performance

Lazar Jovanovic's Best Posts (last 30 days)