Igor Kushnir

B2B operations | Growth consultant. I write breakdowns: what breaks growth and how to fix it | AI + Crypto enthusiast

[Igor Kushnir @TheOnyxis]

I think a lot of managers are having a hard time finding jobs right now for a pretty simple reason.

They know how to manage people.

But companies are starting to look for something more.

A manager today can't just be the person who creates tasks, checks deadlines and goes to meetings.

You need to understand how the actual business works.

Where AI can help.

Where it can't.

What processes can be automated.

What still needs a human.

And most importantly, how all of this actually improves the company.

For example, saying:

“I know ChatGPT and can manage a team using AI”

doesn't really mean much.

But saying:

“I found a process that takes the team 30 hours every week, automated part of it, kept human control where it matters, and reduced the workload to 10 hours”

is a completely different conversation.

I think this is where the role of a manager is changing.

You don't necessarily need to become a developer.

But you probably need to understand technology well enough to connect people, processes, AI and business goals.

The interesting part is that companies don't just need “AI people”.

They need people who understand where AI actually makes sense.

And I think that's becoming one of the most valuable skills for managers.

[Igor Kushnir @TheOnyxis]

Looking for a job in Web3 sounds pretty easy.

You open X, Telegram, LinkedIn, search for “Web3 jobs” and start applying.

But after doing this for a while, you start noticing something:

A lot of these jobs are… questionable.

For example, you find a nice-looking position:

$2k–4k/month
Remote
Web3 startup
Flexible schedule
Sounds good.
You open the website.
Looks good.

You check X.

Looks active.

Then the “recruiter” tells you:

“Just connect your wallet for verification.”

🚩

That's when you close the tab.

And honestly, this changed how I search for Web3 jobs.

I don't really start with the vacancy anymore.

I start with the company.

Who is behind it?
Who works there?
What are they actually building?
Is there any activity outside their own website?
Does the recruiter actually belong to the company?

Because in Web3, a good-looking job post doesn't mean much.

The funny part is that sometimes spending 10 minutes researching the company can save you hours of applying — or much worse.

So now my process is basically:
Company → people → product → activity → vacancy → application.
Not:
“$3k Web3 job — APPLY” 😂
The Web3 job market is interesting.

You just have to learn how to separate opportunities from traps.

[Igor Kushnir @TheOnyxis]

@Atlas3DAI It’s a really cool tool for creating 3D characters.

[Igor Kushnir @TheOnyxis]

I recently played around with DraggableScrollableSheet in Flutter.

At first I thought — okay, just another bottom sheet.

But then I started thinking about where it actually makes sense.

Imagine a crypto app.

You open a transaction and see just the basic info.

Then you swipe up.

Suddenly you have the full transaction history, charts, fees and details — without leaving the screen.

And you can take the same idea even further.

For example, an AI agent dashboard:

collapsed → current task + status
expanded → logs, tools, actions and history
One screen, but two completely different levels of information.

Honestly, I think components like this are more interesting when you stop using them exactly as they were intended.

[Igor Kushnir @TheOnyxis]

@trikcode I think AI slop is inevitable on your page. Sadly

[Igor Kushnir @TheOnyxis]

Most authors write about what interests them

Meanwhile the market is looking for something else entirely.

Someone sits there thinking: “this topic feels close to me, so people will like it too.”
Then they’re surprised the reach is dead, nobody takes the product, and ads disappear into the void.

If the audience is still small — that’s a death sentence.
You can’t afford posts “for the soul.”
The market doesn’t know you yet. It doesn’t care what you’re into. It needs its own question answered.

How it looks in real life:
The author writes “my thoughts on…”
And people in search and in comments are asking:
“what if my situation is different?”
“why isn’t this working for me?”
“is this even worth doing?”
“what’s the first mistake everyone makes?”

That’s demand.
Not a pretty topic. Not inspiration. An unanswered question.

And then it gets funny: when they need reach or they need to sell, they make even more content about themselves.
As if talking louder about their dream will make the market start funding it.
It won’t.

If the product or the post isn’t about the market, you stay the main sponsor. With money, time, nerves.

My take: the market already wrote the brief. Just in messy language. Most authors don’t read it because they’re busy with themselves.

My case: before a post, spend 20 minutes looking not at your own feed, but at what people type into search, what they argue about in comments under someone else’s hits, which questions show up in different places. If the question repeats — that’s the topic. If you invented the topic and nobody’s asking — that’s a diary.

I’ve already started working this way. You can follow the results.

This post was not written by AI — just corrected 😉

My socials
LinkedIn: https://t.co/3174l6NNH9
Telegram: https://t.co/MSEX4PPPfN

[Igor Kushnir @TheOnyxis]

Yes, it’s really good that Anthropic takes safety so seriously and prevents Claude from being used for harm. That’s important.
But sometimes those same safeguards make it quite difficult to have normal conversations on certain topics, even when the intent is completely peaceful and useful. It would be nice if the system were a bit better at distinguishing real misuse from ordinary questions asked for good purposes

[Igor Kushnir @TheOnyxis]

While everyone argues about chatbots, Beijing already wrapped the second “robot Olympics.”

World Humanoid Robot Games 2026, the Ice Ribbon: 666 teams, 2,056 robots, 16 countries, 51 events. In one year the field grew about 4×. Official site: https://t.co/SOKX5EBM7F
The provocation: Tiangong Ultra ran 100m in 8.64s — faster than Bolt’s record. So humanoids already beat humans?

The resolution: the numbers are loud, the human comparison is messy (different rules and conditions). What matters more: in a year the 100m dropped from ~21.5s to under 9, more real-world scenarios, tougher autonomy requirements. Medal leader — AGIBOT, then Tien Kung. It’s a show and a testbed — but the signal on hardware and control is already clear.

My take: the year-over-year progress is hard to miss. I think everyone should already be sketching where, in a few years, “their” robot might show up in their own domains — because access looks like it’s speeding up.

My case: you can follow the companies and people from these Games on socials. If you’ve got an idea — reach out and build something together while the market is still forming.

One more move: don’t only watch the medals. Watch what teams post after the Games — repeatable demos, code, hiring, pilots. That’s where you see who’s shipping a product vs who only ran a pretty sprint.

This post was not written by AI — just corrected 😉

LinkedIn: https://t.co/8Eelqf9AfX:
Telegram: https://t.co/VDLy3IEAkt

[Igor Kushnir @TheOnyxis]

@SlimeX_DeFi ?

[Igor Kushnir @TheOnyxis]

My take on blockchain specialization: why there's no single "best" chain, just different purposes. A deep dive into Ethereum, Solana, BNB, and L2s https://t.co/QjhKaxyes1

[Igor Kushnir @TheOnyxis]

The channel worked. The offer didn't.

A client spent $5,000 on ads.

Got 3 leads.

Said: "The channel doesn't work."

The channel worked.

The offer didn't.

People clicked. Looked. Left.

The ads promised growth.

The website said: "full-service marketing."

Nobody buys "full-service."

They buy a specific pain.

We didn't raise the budget.

We rewrote one line:

not "marketing,"

but "leads from the same traffic in 30 days — or we refund the work."

Leads showed up.

Same budget.

Ads don't create demand.

They speed up what's already there.

If there's no demand — they just burn money faster.

Is your advertising broken

or are you selling something nobody is looking for?

[Igor Kushnir @TheOnyxis]

Maybe your business is already falling behind technology.

You just haven't noticed it yet.

The edge isn't the newest AI tool — it's seeing earlier where it actually helps.

Full post:
https://t.co/yyso7YKsLP

[Jaydeep @_jaydeepkarale]

Building software looked so different just 4 years ago

[Igor Kushnir @TheOnyxis]

Come on, admit it—who else works like this?

[Dilum Sanjaya @DilumSanjaya]

Asked GPT-6 Astra to create a highly detailed, interactive visualization of a V8 engine

The result is pretty impressive

[Igor Kushnir @TheOnyxis]

@RoundtableSpace Neural slopes

[Igor Kushnir @TheOnyxis]

MacBook Pro is the better investment for me. It’s a complete, ready‑to‑use setup (screen, keyboard, trackpad, battery) in one device, so you don’t need to buy extra peripherals. With a lifestyle that requires flexibility and mobility—working from different places, traveling between cities, or just moving around the house—the ability to take your full workstation anywhere is worth the extra cost. Mac mini can be great for price‑to‑performance at a fixed desk, but if it’s your only computer and you need to be mobile, MacBook Pro gives more overall utility

[Igor Kushnir @TheOnyxis]

@LucasAnsel_Cops @Guronnimo This is brilliant, the only thing missing is a system for interaction between visitors, otherwise it's cool.

[Igor Kushnir @TheOnyxis]

@bygregorr @PythonPr finally a sensible answer

[Igor Kushnir @TheOnyxis]

Recently checked my emails for data leaks via @serus_ai

One of my old work emails showed up in 4 dumps. One of them even had a password I used about 6 years ago.

Also checked a couple of friends’ emails (with their permission) — one of them was in a major 2023 service breach.

Simple takeaway: even if you only use a service like this once a year, it’s already useful. At least you know where your data is already floating around and whether it’s time to change passwords or emails.

Have you ever checked your data for leaks? What did you find?