Nick Laptev
@blog.theone.archi
Building theone.archi – the all-in-one hub for architecture artifacts with AI assistance.
ex @IBM, @N26, @Accenture, @thomsonreuters, @Inditex
ex @IBM, @N26, @Accenture, @thomsonreuters, @Inditex
Pinned
Software Architecture | Nick | Substack
This hub is about unique aspects of software architecture that are hard to find or simply not covered in other sources. If well known topics come up, they are always discussed with a strong focus on p...
blog.theone.archi
I write about unique aspects of software architecture that are hard to find or simply not covered in other sources.
All posts ➡️ blog.theone.archi
All posts ➡️ blog.theone.archi
Full post is available here blog.theone.archi/p/the-most-d...
The Most Dangerous Decisions an Architect Makes
An architect makes the most important technical decisions in a company.
blog.theone.archi
February 11, 2026 at 9:39 PM
Full post is available here blog.theone.archi/p/the-most-d...
Architects don’t reduce risk by documenting assumptions.
They reduce risk by testing them.
How do you validate your Big Bang Decisions?
They reduce risk by testing them.
How do you validate your Big Bang Decisions?
February 11, 2026 at 9:39 PM
Architects don’t reduce risk by documenting assumptions.
They reduce risk by testing them.
How do you validate your Big Bang Decisions?
They reduce risk by testing them.
How do you validate your Big Bang Decisions?
Results:
• Core workflows configured and tested
• Custom analytics widget built
• Discovered early that the landing-page widget model wouldn’t scale
Total time: 1 day.
Risk removed.
Stakeholder confidence gained.
• Core workflows configured and tested
• Custom analytics widget built
• Discovered early that the landing-page widget model wouldn’t scale
Total time: 1 day.
Risk removed.
Stakeholder confidence gained.
February 11, 2026 at 9:39 PM
Results:
• Core workflows configured and tested
• Custom analytics widget built
• Discovered early that the landing-page widget model wouldn’t scale
Total time: 1 day.
Risk removed.
Stakeholder confidence gained.
• Core workflows configured and tested
• Custom analytics widget built
• Discovered early that the landing-page widget model wouldn’t scale
Total time: 1 day.
Risk removed.
Stakeholder confidence gained.
Real example:
For a nonprofit project, one Big Bang Decision was: Where will staff and volunteers run core business processes?
Hypotheses:
• AmoCRM can serve as the backbone
• It can be extended for nonprofit needs
I prototyped both.
For a nonprofit project, one Big Bang Decision was: Where will staff and volunteers run core business processes?
Hypotheses:
• AmoCRM can serve as the backbone
• It can be extended for nonprofit needs
I prototyped both.
February 11, 2026 at 9:39 PM
Real example:
For a nonprofit project, one Big Bang Decision was: Where will staff and volunteers run core business processes?
Hypotheses:
• AmoCRM can serve as the backbone
• It can be extended for nonprofit needs
I prototyped both.
For a nonprofit project, one Big Bang Decision was: Where will staff and volunteers run core business processes?
Hypotheses:
• AmoCRM can serve as the backbone
• It can be extended for nonprofit needs
I prototyped both.
“But that takes too long.”
Does it?
With cloud, serverless, and AI-assisted coding, most critical hypotheses can be validated in 4–8 hours.
If a decision can destroy your system, isn’t that worth a day?
Does it?
With cloud, serverless, and AI-assisted coding, most critical hypotheses can be validated in 4–8 hours.
If a decision can destroy your system, isn’t that worth a day?
February 11, 2026 at 9:39 PM
“But that takes too long.”
Does it?
With cloud, serverless, and AI-assisted coding, most critical hypotheses can be validated in 4–8 hours.
If a decision can destroy your system, isn’t that worth a day?
Does it?
With cloud, serverless, and AI-assisted coding, most critical hypotheses can be validated in 4–8 hours.
If a decision can destroy your system, isn’t that worth a day?
“Should the architect really build it?”
Yes. Who else fully owns the risk?
Prototyping forces you to confront reality instead of trusting your intuition.
It grounds the architecture in execution.
Yes. Who else fully owns the risk?
Prototyping forces you to confront reality instead of trusting your intuition.
It grounds the architecture in execution.
February 11, 2026 at 9:39 PM
“Should the architect really build it?”
Yes. Who else fully owns the risk?
Prototyping forces you to confront reality instead of trusting your intuition.
It grounds the architecture in execution.
Yes. Who else fully owns the risk?
Prototyping forces you to confront reality instead of trusting your intuition.
It grounds the architecture in execution.
“What counts as a prototype?”
Anything minimally functional:
• A bash script
• A CLI tool
• A plugin
• A small deployed system
It doesn’t need to be pretty.
It needs to prove something.
Anything minimally functional:
• A bash script
• A CLI tool
• A plugin
• A small deployed system
It doesn’t need to be pretty.
It needs to prove something.
February 11, 2026 at 9:39 PM
“What counts as a prototype?”
Anything minimally functional:
• A bash script
• A CLI tool
• A plugin
• A small deployed system
It doesn’t need to be pretty.
It needs to prove something.
Anything minimally functional:
• A bash script
• A CLI tool
• A plugin
• A small deployed system
It doesn’t need to be pretty.
It needs to prove something.
An invalidated hypothesis is NOT failure.
It’s saved money. Saved time. Saved reputation.
In my experience:
• Experienced architects get ~30% of assumptions wrong
• Inexperienced ones can be wrong up to 80%
Unchecked assumptions are expensive.
It’s saved money. Saved time. Saved reputation.
In my experience:
• Experienced architects get ~30% of assumptions wrong
• Inexperienced ones can be wrong up to 80%
Unchecked assumptions are expensive.
February 11, 2026 at 9:39 PM
An invalidated hypothesis is NOT failure.
It’s saved money. Saved time. Saved reputation.
In my experience:
• Experienced architects get ~30% of assumptions wrong
• Inexperienced ones can be wrong up to 80%
Unchecked assumptions are expensive.
It’s saved money. Saved time. Saved reputation.
In my experience:
• Experienced architects get ~30% of assumptions wrong
• Inexperienced ones can be wrong up to 80%
Unchecked assumptions are expensive.
There’s a better way.
For every critical assumption → build a prototype.
Not slides.
Not diagrams.
Something that actually runs.
Validate before you commit.
For every critical assumption → build a prototype.
Not slides.
Not diagrams.
Something that actually runs.
Validate before you commit.
February 11, 2026 at 9:39 PM
There’s a better way.
For every critical assumption → build a prototype.
Not slides.
Not diagrams.
Something that actually runs.
Validate before you commit.
For every critical assumption → build a prototype.
Not slides.
Not diagrams.
Something that actually runs.
Validate before you commit.
Here’s the common pattern:
– Architect defines assumptions.
– Logs them.
– Finalizes the design.
– Moves on.
– Six months later everything explodes.
Response? “That was an assumption.”
Technically true. Practically useless.
– Architect defines assumptions.
– Logs them.
– Finalizes the design.
– Moves on.
– Six months later everything explodes.
Response? “That was an assumption.”
Technically true. Practically useless.
February 11, 2026 at 9:39 PM
Here’s the common pattern:
– Architect defines assumptions.
– Logs them.
– Finalizes the design.
– Moves on.
– Six months later everything explodes.
Response? “That was an assumption.”
Technically true. Practically useless.
– Architect defines assumptions.
– Logs them.
– Finalizes the design.
– Moves on.
– Six months later everything explodes.
Response? “That was an assumption.”
Technically true. Practically useless.
Big Bang Decisions shape the entire architecture.
If you get one wrong, you don’t just refactor.
You rebuild. Or you fail.
They are high-risk by definition.
If you get one wrong, you don’t just refactor.
You rebuild. Or you fail.
They are high-risk by definition.
February 11, 2026 at 9:39 PM
Big Bang Decisions shape the entire architecture.
If you get one wrong, you don’t just refactor.
You rebuild. Or you fail.
They are high-risk by definition.
If you get one wrong, you don’t just refactor.
You rebuild. Or you fail.
They are high-risk by definition.
The most dangerous decisions a software architect makes.
When designing a system from scratch, you’ll easily make 50+ architectural decisions.
But 3–5 of them determine whether the whole system succeeds or collapses.
I call them Big Bang Decisions 🧵
When designing a system from scratch, you’ll easily make 50+ architectural decisions.
But 3–5 of them determine whether the whole system succeeds or collapses.
I call them Big Bang Decisions 🧵
February 11, 2026 at 9:39 PM
The most dangerous decisions a software architect makes.
When designing a system from scratch, you’ll easily make 50+ architectural decisions.
But 3–5 of them determine whether the whole system succeeds or collapses.
I call them Big Bang Decisions 🧵
When designing a system from scratch, you’ll easily make 50+ architectural decisions.
But 3–5 of them determine whether the whole system succeeds or collapses.
I call them Big Bang Decisions 🧵
The results are published automatically every day here:
t.me/prod_ideas
No hype. No “idea generators”.
Just structured, real-world pain.
Founders, consider this a free database of ideas for your next product 🎁
t.me/prod_ideas
No hype. No “idea generators”.
Just structured, real-world pain.
Founders, consider this a free database of ideas for your next product 🎁
Product Ideas
Generate daily AI based analytics over Reddit posts where people share pain in their business. Made by @nickolay_laptev.
t.me
January 28, 2026 at 9:42 PM
The results are published automatically every day here:
t.me/prod_ideas
No hype. No “idea generators”.
Just structured, real-world pain.
Founders, consider this a free database of ideas for your next product 🎁
t.me/prod_ideas
No hype. No “idea generators”.
Just structured, real-world pain.
Founders, consider this a free database of ideas for your next product 🎁
For each problem, the AI answers:
– Is this actually a real problem?
– What’s the root cause?
– How hard is it to solve?
– How popular is it?
Building for ultra-niche audiences is a very specific kind of pleasure 🙂
– Is this actually a real problem?
– What’s the root cause?
– How hard is it to solve?
– How popular is it?
Building for ultra-niche audiences is a very specific kind of pleasure 🙂
January 28, 2026 at 9:42 PM
For each problem, the AI answers:
– Is this actually a real problem?
– What’s the root cause?
– How hard is it to solve?
– How popular is it?
Building for ultra-niche audiences is a very specific kind of pleasure 🙂
– Is this actually a real problem?
– What’s the root cause?
– How hard is it to solve?
– How popular is it?
Building for ultra-niche audiences is a very specific kind of pleasure 🙂
So I decided to do this:
Collect real user pain from Reddit and run AI analysis on top of those complaints.
Not opinions.
Not guesses.
Actual problems people already care about.
Collect real user pain from Reddit and run AI analysis on top of those complaints.
Not opinions.
Not guesses.
Actual problems people already care about.
January 28, 2026 at 9:42 PM
So I decided to do this:
Collect real user pain from Reddit and run AI analysis on top of those complaints.
Not opinions.
Not guesses.
Actual problems people already care about.
Collect real user pain from Reddit and run AI analysis on top of those complaints.
Not opinions.
Not guesses.
Actual problems people already care about.
They complain online.
A lot.
And for some reason, they especially love doing this on Reddit:
– Ugly design.
– Constant bugs.
– Random bans.
– Questionable product decisions.
Yet people still go there to share what’s bothering them.
A lot.
And for some reason, they especially love doing this on Reddit:
– Ugly design.
– Constant bugs.
– Random bans.
– Questionable product decisions.
Yet people still go there to share what’s bothering them.
January 28, 2026 at 9:42 PM
They complain online.
A lot.
And for some reason, they especially love doing this on Reddit:
– Ugly design.
– Constant bugs.
– Random bans.
– Questionable product decisions.
Yet people still go there to share what’s bothering them.
A lot.
And for some reason, they especially love doing this on Reddit:
– Ugly design.
– Constant bugs.
– Random bans.
– Questionable product decisions.
Yet people still go there to share what’s bothering them.
The classic answer is simple: “Talk to people.”
That still works. But we live in the 21st century.
When something truly hurts, people don’t stay quiet about it anymore.
That still works. But we live in the 21st century.
When something truly hurts, people don’t stay quiet about it anymore.
January 28, 2026 at 9:42 PM
The classic answer is simple: “Talk to people.”
That still works. But we live in the 21st century.
When something truly hurts, people don’t stay quiet about it anymore.
That still works. But we live in the 21st century.
When something truly hurts, people don’t stay quiet about it anymore.
Ideas for new products 🧵
A successful product solves real problems people have.
Not imaginary problems invented by the founder 😁
So the real question is:
How do you actually find real problems?
A successful product solves real problems people have.
Not imaginary problems invented by the founder 😁
So the real question is:
How do you actually find real problems?
January 28, 2026 at 9:42 PM
Ideas for new products 🧵
A successful product solves real problems people have.
Not imaginary problems invented by the founder 😁
So the real question is:
How do you actually find real problems?
A successful product solves real problems people have.
Not imaginary problems invented by the founder 😁
So the real question is:
How do you actually find real problems?
Full post is here blog.theone.archi/p/viral-tren...
Viral Trends and Pure Idiocy 😁
Some time ago, a tool called Clawdbot popped up.
blog.theone.archi
January 26, 2026 at 11:33 AM
Full post is here blog.theone.archi/p/viral-tren...
Viral Trends and Pure Idiocy 😁
Clawdbot is a useful messenger UI for AI. Not magic.
Yet people:
* buy Mac minis
* burn money on APIs
* give bots passwords
* delegate life to agents with prompt injection
The tool is fine. The hype is the problem.
Clawdbot is a useful messenger UI for AI. Not magic.
Yet people:
* buy Mac minis
* burn money on APIs
* give bots passwords
* delegate life to agents with prompt injection
The tool is fine. The hype is the problem.
January 26, 2026 at 11:33 AM
Viral Trends and Pure Idiocy 😁
Clawdbot is a useful messenger UI for AI. Not magic.
Yet people:
* buy Mac minis
* burn money on APIs
* give bots passwords
* delegate life to agents with prompt injection
The tool is fine. The hype is the problem.
Clawdbot is a useful messenger UI for AI. Not magic.
Yet people:
* buy Mac minis
* burn money on APIs
* give bots passwords
* delegate life to agents with prompt injection
The tool is fine. The hype is the problem.
Full story ➡️ blog.theone.archi/p/when-a-com...
When a company suddenly decides to fight for high productivity 🤡
My former colleague, now a Director of Engineering and overall a great guy, Fabiano, did a breakdown of a recent case in Brazil where a large bank fired 1,000 employees.
blog.theone.archi
January 16, 2026 at 12:47 PM
Full story ➡️ blog.theone.archi/p/when-a-com...
A bank fired 1,000 people for “low productivity” based on mouse clicks and keystrokes 🤡
Some of them were promoted for high performance a month earlier.
When you can’t measure real impact, you measure nonsense.
Then people start gaming the system.
Thoughts?
Some of them were promoted for high performance a month earlier.
When you can’t measure real impact, you measure nonsense.
Then people start gaming the system.
Thoughts?
January 16, 2026 at 12:47 PM
A bank fired 1,000 people for “low productivity” based on mouse clicks and keystrokes 🤡
Some of them were promoted for high performance a month earlier.
When you can’t measure real impact, you measure nonsense.
Then people start gaming the system.
Thoughts?
Some of them were promoted for high performance a month earlier.
When you can’t measure real impact, you measure nonsense.
Then people start gaming the system.
Thoughts?
I write about unique aspects of software architecture that are hard to find or simply not covered in other sources.
All posts ➡️ blog.theone.archi
All posts ➡️ blog.theone.archi
Software Architecture | Nick | Substack
This hub is about unique aspects of software architecture that are hard to find or simply not covered in other sources. If well known topics come up, they are always discussed with a strong focus on p...
blog.theone.archi
January 14, 2026 at 12:55 PM
I write about unique aspects of software architecture that are hard to find or simply not covered in other sources.
All posts ➡️ blog.theone.archi
All posts ➡️ blog.theone.archi