Pawel Brodzinski

Vibe-coded app. Authorization.

"Add localhost:3000/auth/linkedi... to authorized redirect URLs"

It is "works well on my machine" of the AI era :D

Just a couple years ago, it made the news that CD Projekt RED (of Witcher / Cyberpunk fame) started adopting this thing called Agile.

In 2023.

Give them 2 more decades, and they will catch up with modern management and Lean Startup, too.

I've recently seen someone bragging about a badge from OpenAI about how many tokens they used.

My reaction? Well, you burned energy that would be enough to supply a few hundred households for a year. Con-fucking-gratulations!

I recently referenced Brooks' Mythical Man-Month. Already in 1975, he pointed out the issues with productivity measures such as Lines of Code (even if he didn't have much better tools).

Half a century later, we're not *that* far from LoC today...

"Unlike 1999, tech’s profits justify the premium."

Are these profits coming from AI? Save for hardware vendors, which AI wave made fly, what profits are we talking about?

And what happens to Nvidia when it's suddenly clear that OpenAI isn't on a path to profitability?

substack.com/inbox/post/1...

Drama Game: One Word Story - Beat by Beat Press Type:  Warm-up. Purpose: To work as a team. To work on focus. Materials: A big enough space for the entire class to sit in a circle comfortably. Procedure:  Players sit in a circle. One person says a ...

Just thought about it: an LLM is playing One Word Story except all the players are the LLM.

bbbpress.com/2013/01/one-...

Hype wave versus reality.

Upon seeing the chart, I was like:
* Your ARR is not annual (can't be yet)
* Nor is it recurring (can't know yet)

After 10s research (literally clicking the links):
* It's also fake
* And scam

Check your sources, people!

"We haven't yet shown a single dollar of profit, but we're already too big to fail."

Welcome to 2025.

What do you not understand?

One of the most valuable aspects of an organizational culture is what I call an experimentation mindset. It's about being open and willing to continuously try things and learn from the process. O... One of the most valuable aspects of an organizational culture is what I call an experimentation mindset. It's about being open and willing to continuously try things and learn from the process. Or in...

Here's my little experiment on how proliferated LinkedIn is with automatic AI bots. I would appreciate sharing and/or reacting to it to increase its range:

www.linkedin.com/feed/update/...

Thank you in advance.

@kentbeck.com's post (substack.com/inbox/post/1...) reminded me about my own "disastrous failure" story.

In the end, our clients just ended up with a bit of extra work. Nothing too serious.

That didn't age overly well.

It still somehow escapes me how these English developers build successful products.

For years, statistical forecasting was the way to go for me when someone asked for an estimate.

Then I got back to simple linear approximation.

TL;DR: It's much quicker. And good enough.

www.linkedin.com/feed/update/...

Lovable’s ARR is Vanity Metric 2.0 Lovable's claim to be the fastest growing startup ever is based on their Annual Recurring Revenue traction. In their case, however, ARR is basically a vanity metric.

My not-so-love letter to Lovable's stellar growth.

TL;DR: Lovable's reported growth is a classic vanity metric, and it won't last.

pawelbrodzinski.substack.com/p/lovables-a...

Despite our trade being reformatted, I am optimistic that in the long run, we aren't going to be significantly worse off.

And I can't say that I'm that optimistic about some other trades.

Thanks for an extensive reply. I misread the original message a bit, as if it were joining the choir of the disadvantaged. While it was more a voice of empathy.

At a risk of being controversial: IT should be the last to complain.

We've been a privileged class for 30+ years, and we still have a disproportionately advantageous position by actually developing lots of AI solutions.

Also, as for current state of tech, we'll still be needed.

Code From a 2015 BlackHat presentation

Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it. - Brian Kernighan

Linear announced "the best feature ever" which is Slack integration that automatically adds a work item based on AI interpreting the conversation.

Well, so much for judgment in product development.

That's a totally new level of AI slop. Work generated by AI for humans :)

Vanity metric: a metric that makes you look good, but doesn't necessarily help to understand company performance or inform its strategy.

ARR should thus be a vanity metric.

It tells nothing about:
* ramping AI infrastructure costs (per user)
* hype-wave effect on future revenue predictions

You Don’t Rise to the Level of Your Goals, You Flow to the Level of Your Systems Why great teams don’t chase bigger goals, but build better systems that make ambition sustainable.

"Goals are tidy. Systems are messy."

We don't fix productivity by sprinkling OKRs at our work.

It won't do much unless we agree on the basics:
What does "in progress" mean?
How do we choose the next most important thing?
What's the most desired outcome of our work?

substack.com/home/post/p-...

Future Plateau: Neat Example of a Compound Metric Future Plateau is a practical example of a forward-looking, compound metric, which allows for a neat what-if analysis.

"ARR tends to be nothing more than just an appealing lie."

What's reported as annual and recurring these days, surprisingly rarely is either.

substack.com/home/post/p-...

Why AI cannot simulate your customers behaviour "Founders would literally rather boil the ocean than talk to customers."

“AI Models Can Now Predict What Customers Would Buy—Almost Like Humans”

Now, can they?

TL;DR: Nope.

substack.com/inbox/post/1...

Trust Networks as Antidote to AI Slop - Pawel Brodzinski on Leadership in Technology In response to AI slop we will increasingly rely on trust networks in professional dealings. Trust will be the new business currency.

So, how do we navigate in a landscape where we can't trust that any bit of content is:
a) created by a human
b) not fake
?

Trust networks. Something we've been doing for millennia.

As a bonus, it's also a damn good antidote for AI slop.

brodzinski.com/2025/10/trus...

Code review karma: the better code reviews you give, the better ones you receive.

That and more stats behind code reviews): substack.com/inbox/post/1...

Find 5 differences.

The top picture is from Musk's tweet mocking the AWS outage.
The bottom is from the original CNBC article.

(And it's not to defend AWS or anything.)

When you're drinking from a firehose, the performance limit in that situation isn't the firehose.

Candidates mass applying with AI.
Companies mass rejecting with AI.
(Not so new) hacks to trick the AI tools emerge.

Both hiring and looking for a job is sure fun these days.

Is it only me, or is it not sustainable?

Source: substack.com/inbox/post/1...

XKCD explains the Cloud

xkcd.com/908/
This classic XKCD cartoon seems...relevant today

Recruitment in 2025.

AI has helped everyone so much! High fives, everyone!

When they ask "What the hell does it even mean to fail a sprint and why would we care?" that's a sign of even bigger maturity ;)

Sprint success/failure is entirely artificial.

You can fail a sprint and do the right thing/deliver a lot of value. And vice versa.