Unconventional Leadership Enthusiast | Creator and propagator of the "5E's of Leadership" | Empowering Leaders to Transform & Thrive in the Age of Change

When leaders face tough situations — turnarounds, post-acquisition chaos, leadership breakdowns, toxic culture, execution failures — they don’t need advice or frameworks.
They need someone who can step in, take ownership, and deliver results.

That’s what I do.

I bring 20+ years of senior HR and organizational leadership experience across EMEA, with 10+ large-scale transformation assignments. I embed with executive teams, stabilize situations fast (3–6 months), and build sustainable leadership and organizational capability over 12–18 months.

I work at the intersection of leadership, decision-making, and organizational systems — where strong individuals often struggle to operate as a coherent whole under pressure.

⸻

Selected impact:

→ German manufacturing business in crisis
Leadership vacuum, toxic culture, massive talent exodus.
Turned €10M annual losses into breakeven within 12 months.

→ Three Italian plants operating as warring factions
No alignment, no shared priorities.
Built cross-plant leadership alignment; European product sales increased by ~25%.

→ Young leadership team blocked by entrenched resistance
Coached and supported them through a full transformation.
Order execution speed improved by 30%; complete business model turnaround within one year.

→ Severe labor availability crisis across Eastern Europe
Attrition reduced from 25% to 8%; seasonal staffing stabilized through leadership and HR mobilization.

→ Quality function collapse after global product recall
Rebuilt the function from scratch, replaced half of leadership.
All critical issues resolved within 12 months; quality records broken year after year since.

→ Commercial organization with toxic culture and recurring losses
Transformed into the fastest-growing commercial team in Europe within 18 months.

⸻

How I work

Beyond classic interim leadership, I increasingly use facilitated business simulations to make leadership and decision-making dynamics visible under pressure.
They help executive teams see — within hours — patterns that often remain hidden in meetings for years.

This is not training and not theory.
It’s a way to surface how decisions interact, where responsibility concentrates or disappears, and why strong teams sometimes underperform as a system.

⸻

Where I typically engage

• Crisis stabilization
• Post-acquisition integration
• Leadership transitions
• Culture and execution transformation

Scope: EMEA.

⸻

Available for Interim CHRO / HR Director assignments and executive-level transformation work.
High travel is not an issue.


Mikhail Lvovskii

I used to introduce myself as an HR Business Partner.

Then a plant manager told me: I don't need a partner. I need someone who tells me what I don't want to hear.

That stuck.

See, "partner" sounds strategic. It sounds like you're in the room, working together, nodding along.

But in the real world, when a business is in trouble,
agreement is the last thing that moves the needle.

Leaders don't need another person cosigning their decisions.
They need someone who points out what everyone else is quietly ignoring.
Someone who's willing to be unpopular.
Someone who calls out the elephant, every single time.

That's what makes you valuable.
Not sitting at the table.
Not your seat on the org chart.

The best HR people aren't just partners.
They're mirrors. The kind that shows the full, messy picture,no matter who's looking.

undefined undefined

Happy New 2006 Year to everyone!
Best wishes to you and your closed ones!

Certified Transformational Coach

Global Professional in Human Resources (GPHR)

Change Management. Level 4. Change Trainer.

Coaching for Performance. Train the Trainer

Continuous Improvement and Lean Management

EQ 360 degrees feed-back coach

Efficient methods of recruitment and headhunting

Electrolux Leader

Electrolux Next Level Leader

HR Academy

Job Evaluation: Mercer's methodology

People management skills

People management training. Train the trainer

Preparation to GPHR certification

Samsung Leadership program

StoreWars business simulation

Storytelling

Interim HR leadership and transformation execution for organizations facing crisis, transition, or underperformance.

Short- to mid-term Organizational interventions to the teams with dysfunctions, in order to build high the basis for high performance.

Selected projects:

Pharma (multinational, CEE) – Reset the top team, introduced ambitious visioning and planning cadence, strengthened collaboration → positioned business for accelerated market growth.

Global Electronics producer – Advised on HR operating model redesign, aligning service delivery with business strategy → increased ownership by line managers, freed HR from admin.

Industrial Plant (Germany/Poland border, 900 employees) – Installed accountability ladder and reset leadership culture → productivity lifted by 20% in 3 months.

Law Firm (Poland, 400 employees) – Partnered with new HR Director after failed engagement survey → leadership reset, trust rebuilt, attrition slowed within 6 months.

Founder and Principal Consultant

Helping companies overcome challenges in the area of people, leadership and culture.
Helping to run organizational transformation: design, implementation, change management.
Transforming HR organizations to be true business partners and move them towards business drivers.

Chief Transformational Officer

Built HR function from scratch for international logistics company: strategy through execution, all processes, leadership team across 15 countries.

Led three major turnaround initiatives:

German Entity Turnaround (2023-2024)
Situation: German manufacturing entity in crisis — leadership vacuum, toxic culture, talent exodus. Running consistent losses (-10M EUR annually) despite major technology investment.
Action: Led leadership and culture reset. Addressed talent crisis through targeted interventions. Aligned leadership team around performance recovery. Reviewed and streamlined key functional processes.
Result: From -10M EUR annual losses to breakeven within 1 year. Organization stabilized, leadership aligned, underperformers exited.

Italian Operations Integration (2023-2024)
Situation: Three Italian plants operating as warring factions — fighting internally and resisting collaboration with European operations.
Action: Drove cross-plant alignment initiatives. Broke down silos between Italian entities and broader European organization. Built collaborative leadership culture.
Result: Italian product sales in Europe increased ~25%. Unified operations, established productive collaboration. Refreshed Italian management at top 2 layers.

Lithuania Leadership Transition (2022-2024)
Situation: Young leadership team installed after leadership reset, struggling to drive change in entrenched old-guard organization.
Action: Coached and supported new leadership team through transformation. Helped build credibility and change capability. Addressed resistance, aligned organization around new direction.
Result: Orders execution speed improved 30%. Complete business model turnaround from imports to local production within 1 year. Annual growth 15%.

European operations sold to strategic buyer in 2024 following successful stabilization.

Chief Human Resources Officer

Progressive HR leadership across EMEA Industrial Operations, with four major turnaround/transformation cases:

Quality Function Crisis Recovery (2018-2019)
Situation: Global quality scandal — mass product recall (dishwashers) causing massive financial losses and reputational damage. Quality function neglected for years: incompetent team, weak culture, no credibility.
Action: Partnered with new Quality leader to rebuild from ground up. Replaced more than half of leadership team. Redefined function's mission. Recruited high-caliber talent. Coached leader through confidence gaps.
Result: All critical quality issues resolved within 12 months. Historical quality records broken every subsequent year. Function transformed from liability to competitive advantage.

East Europe Manufacturing Labor Crisis (2018-2019)
Situation: By season 2018: zero applications for seasonal workers, significant drop in permanent worker inflow. No one wanted to join Electrolux factories.
Action: Gathered EMEA Manufacturing leadership to develop comprehensive solution. Introduced migrant workers approach despite strong stakeholder resistance.
Result: Decreased workers' attrition from 25% to 8% within months. Employee engagement improved 22 percentage points in Eastern Europe. Season 2018 sdtafed fully. YoY productivity and OEE targets over-achieved for first time in Eastern Europe.

Supply Chain Centralization (2014-2016)
Situation: Major organizational restructuring — centralizing Supply Chain operations across EMEA. Experienced plant managers losing power. Strong resistance, lack of trust, leadership vacuum.
Action: Deployed as embedded change agent. Bridged gaps between central leadership and local plant management. Ensured functioning of new structure during transition.
Result: New centralized organization successfully established and operational. Achieved strategic targets for product delivery at improving cost. New SC organization gained trust and demonstrated advanced capabilities.

HR Director Industrial Operations EMEA (from 2019 - Europe)

Commercial Organization Turnaround
Situation: Dysfunctional commercial organization with toxic culture, chronic underperformance, consistent losses. Managing Director terminated in week two of my arrival.
Action: Kept business running for 4 months while assessing team. Personally removed most toxic individuals. Rebuilt leadership culture around accountability. Supported onboarding of new MD and completed organizational reset.
Result: Transformed from loss-making unit to fastest-growing commercial team in Europe. Delivered significant sales growth and profit turnaround within 18 months. Built self-steering team capable of sustaining performance.

Regional HR Director (CIS counties + Baltics)

HR leadership for industrial gas operations including oversight of two M&A integration projects. Managed complex stakeholder environment, aligned cultures and systems across acquired entities while maintaining operational stability.
Assured stable business functioning along severe Financial crisis of 2008-2009, forcing mass redudancies and business restructuring. While making sure 2 entities acquired pre-crisis were integrating into the business wihtout performance loss.

HR Director

Regional HR leadership across CIS markets for consumer electronics business.
Establishing HR system from scratch along shaping new Global region in CIS.

CIS Region HR Manager

Post-merger integration of Business Consulting practice into IBM structure. Aligning processes, leadership practices and culture.

HR Manager. Business Consulting Services department (integration project)

Recruitment and development support for the pilot project for the emerging business model of dedicated software development centres.

Recruitment Manager

Interim HR Director/CHRO | Turnaround & Transformation Specialist | I stabilize broken organizations and rebuild dysfunctional teams

English

German

Polish

Russian

SHRM

Interim HR Director/CHRO | Turnaround & Transformation Specialist | Helping leaders succeed when context is tough

A factory was bleeding money.

Everyone expected layoffs.

I started with something different.

I asked the shift supervisors what they'd do if they ran the place.

The answers were sharper than anything the leadership team had cooked up.

No management speak. No theory.

They named the broken machines, the wasted steps, and the dumb rules that slowed down every shift.

They didn't need a consultant or a strategy deck.

They needed someone to ask, listen, and act.

The people closest to the work had been watching the real problems for years. Nobody had bothered to ask.

The turnaround kicked off the day leadership stopped pretending they had all the answers.

You want a transformation? 

Before you start buying frameworks and hiring experts, try something radical.

Ask the people staring at the mess every day how they'd fix it.

Then actually listen.

She was one of our best.

On her last day, I asked why she was really leaving.

She said, "I've been leaving for two years. No one noticed."

That's the part no one wants to hear.

By the time someone resigns, the decision was made months ago.

You can track the trail if you bother to look.

First comes the silence in meetings. 
Then the suggestions that go nowhere.
After that, the missed promotion, or no feedback at all.

It's a slow fade.

Most companies don't see it, or pretend not to.

Then they schedule the exit interview, hoping for some honest insight.

But by then, it's just an autopsy.

The real work is upstream. If you want blunt truth, ask before people disengage. 

Stay interviews. Candid check-ins. Not the end, the middle.

Ask early. Ask often. Or get comfortable with good people silently slipping out the door.

You choose which kind of interview you prefer.

Forget teaching.

The fastest way to develop a team isn't to lecture, train, or hand them a new model.

It's to throw them into a real situation that reveals their default patterns, good and bad.

No slides. No expert diagnosis. Just a mirror.

Let them see the gap for themselves.

After 25 years of running all the types of workshops and interventions, I don't believe in top-down development.

People don't shift because someone tells them how.

They shift when they see what they're actually doing, together, in real time.

That discomfort? That clarity? It does the heavy lifting.

The hard part: engineering that moment without artifice or theater.

That's the edge I've been focused on. More on the method soon.

No alternative text description for this image

There are two types of team building.

One asks, "Do we like each other?"
That's escape rooms, cooking classes, drinks after work.
Easy to buy, easy to deliver.
Zero impact on Monday morning.

The other asks, "How do we function when it matters?"
Now you're talking resource constraints, time pressure, clashing priorities.
You put the team in the fire. Watch what surfaces.
It's messy. It's uncomfortable. Most companies run the other way.

After 25 years, I've seen the same pattern.
- Teams that choose comfort stay stuck.
- Teams that choose the mirror, the hard look at reality, break through every time.

You want transformation?
Skip the trust falls.
Choose the mirror.

Networking w swiecie przedsiębiorców jest niezbędny.
Ale jak być jeżeli nie jesteś "stąd"?

Przeprowadziłem się do Wrocławia, no i do Polski, ponad 12 łat temu, po 37 łatach spędzonych w rodzinnym kraju i miejście.

Tak się stało ze długie łata skupiłem się na karjerze korporacyjnej tego pracodawcy który mi tutaj przywiózł. I nie myśliałem że poza ramami tego pracodawcy też jest życie. Nie budowałem sieci.

Tak się wydarzyło że 5 łat temu stwierdziłem że chcę kontynuować swoją drogę samodzielnie, prowadzić własny biznes. 
Ale się okazało że network który miałem był słaby. Jakość sieci po moim ocenam bardzo niska. Większość wartościowych kontaktów to jedna firma - mój były pracodawca.

Probowałem rózne sposoby, ale włąsnie nic nie działa lepiej niż spotkanie tet-a-tet i dzielenie się wiedzą.

Dzięki Martyna Lempert, z której się poznaliśmy z połtora tygodni chyba, już po raz drugi biorę udział w spotkaniu kameralnym przedsiębiorców Wrocławia. I to własnie jest wartość ludzkich relacji - poradzić, skierować, pomóc. 
Dzięki Martyna jeszcze raz!

O wydarzeniu:
Serial networkers Meetup - Wrocław!✅ 

Data: 6 marca 2026
Miejsce: Ace of Space Renoma, Świdnicka 40, Wrocław.
Czas: 16:00-21:00

Bilety: https://lnkd.in/drptwwBB

Team building is HR's polite way of avoiding the real conversation.

A CEO once told me: "The team isn't working. Let's do an offsite."

We did the offsite. Escape room. Dinner. Drinks.

Monday morning, same dysfunction.

Because the problem wasn't that they didn't like each other.
The problem was how they made decisions together — and nobody wanted to name it.

Team building that entertains is easy.
Team building that reveals is uncomfortable.

Guess which one gets budget approval?

€50,000 blown on a leadership offsite.

Three months later, same problems.

Great facilitator. Gorgeous venue. Everyone left happy.

Nothing changed.

The offsite treated symptoms, not systems.
For a week, everyone played nice. Then reality snapped back.

One person still hogged decisions.
Old conflicts, still swept under the rug.
Alignment, still fake.

Feeling good together doesn't fix teams.
Seeing, then fixing, how a team actually works does. 

Most team building is a vacation from dysfunction.
Real development? Brutal. It's a mirror, not a retreat.

HR keeps buying what's easy to explain.

Quiz night? That one sells itself. It's cheap. Everyone joins in. No heavy lifting.

But try selling a simulation that shows which leaders choke, stall, or step up in real decisions.

Suddenly, it's too hard to pitch. The CFO gives you a blank stare.

So HR grabs the thing with the simple story. Entertainment wins. Real change loses.

Then they wonder why nothing sticks. Why leaders keep repeating the same mistakes.

The real blocker in corporate learning isn't a lack of ideas. It's a fear of anything that can't be explained in a single line.

If your solution is neat and fun, it lives.
If it reveals hard truths, it dies in the meeting room.

Real transformation isn't simple. The pitch shouldn't be either.

The best assessment I ever saw wasn't an assessment.

I used to roll my eyes at gamified anything. Always felt like a gimmick.

Then I watched a senior leadership team dive into a business simulation.

No assigned roles. No scripts. No "correct" answers.

Within an hour, everything flipped:

- The quiet one? Turned into the team's strategist. 
- The confident one? Tanked the business with reckless bets. 
- The supposed collaborator? Bulldozed every conversation.

They stopped performing. They just played.

That's when the real dynamics showed up.

You want to know how someone leads? 
Stop calling it an assessment.

Let them forget they're being watched. 
And see who actually shows up.

I watched 12 teams solve the same problem, with the same rules and the same time.

Not a single team did it the same way.

But some patterns stood out.

 Teams with one loud voice moved fast. They also got the worst results.

Teams that "wasted" time arguing every angle? They made fewer mistakes.

And when someone asked, "Wait, what are we optimizing for?", that team usually crushed the ones who just rolled up their sleeves and started working.

The kicker?
Teams that thought they nailed it weren't always the ones who actually did.

Especially under pressure, confidence and competence are two different signals.

If your team moves fast and sounds sure of themselves, you might want to check what they're actually delivering.

The CHRO wanted a succession plan.

I only asked one question:
Who on your team would you actually bet on?

Silence.

Not because they didn't know.
Because saying it out loud means facing the reality.

Most succession planning is corporate theater.
- Map names to boxes.
- Run assessments.
- Draft development plans.
Nothing happens.

The real questions never make the agenda:
- Who's truly ready?
- Who never will be?
- Who's here just because it's easy to keep them?

HR builds frameworks.
Leadership dodges decisions.

If your succession plan hasn't changed in six months,
 it's not a plan.
It's paperwork.

Sent hundreds of managers to leadership training.

Every feedback form was glowing.
"Eye-opening." "Very useful." "Will apply immediately."

But come Monday, nothing changed.

Took me years to see the gap.

Training tells you what good leadership looks like.
It never shows what your leadership actually is.

There's a world of difference between:
- Knowing you should listen more
- And watching yourself interrupt three people in a row when things get tight.

The first is advice.
The second is a mirror.

Most companies spend a fortune on advice.
Almost none bother with mirrors.

That's why so many "trained" managers never improve.

Most leadership teams think their collaboration is strong.

Surveys will back that up.

But watch them try to split a shrinking budget in real time.

That's where the truth shows up.

I once ran a survey for a leadership team. 90 percent rated their collaboration as strong or very strong.

Then I watched them actually work together.

It took 30 minutes for two people to bulldoze every decision. One person checked out completely. The rest held their ground but never moved the conversation forward.

Nobody lied on the survey. They honestly thought they were a high-performing team.

But self-perception is not data.

If you want to know how your team works, stop asking. Start watching.

Observed behavior is the only data that counts.

We ran an engagement survey. The results came back brutal. 
The CEO's first response: "Can we adjust the questions for next year?"

That right there is the difference.

Organizations that want to improve ask, "What do we do with this?"

Organizations that want to look good ask, "How do we explain this?"

It's not the survey that's broken. It's the instinct to flinch at bad news.

If you're not ready to act on the truth, don't ask for it.

A leadership team spent three months debating a senior hire.

In the end, they picked the safest candidate, someone exactly like them.

Here's what most leaders won't admit, even to themselves.

Hiring decisions expose what you actually value, not what you claim to value.

Every time someone gets labeled a "culture fit," take a closer look. Sometimes it's alignment. More often, it's just comfort.

You can paste values on the wall and talk about innovation, but the real test is in the room when a decision needs to be made.

Your hires are your culture.

Look at your last five. What did you really optimize for?

Be honest. That's the only way things change.

The worst team conflict is silent.

I've seen leadership teams who brag about "alignment." No debates. Quick decisions. Lots of nodding.

Looks smooth. Until you see the mess underneath.

Three strategies pulling in opposite directions. Secret alliances. One person overloaded while the rest jockey for status.

The show of harmony was just that, a show.

Healthy teams fight in the room, then walk out aligned. 
Dysfunctional teams nod in the room and undercut each other later.

If your meetings never get heated, don't kid yourself. That's not alignment.
 
That's avoidance.

Every meeting was polite. Every decision was unanimous.

And the company was slowly dying.

Here's the hard truth. If your leadership team never argues, you don't have trust. You have silence.

Real teams fight. They call out bad ideas before they turn into layoffs. They push each other, because they care more about the business than about looking agreeable.

When you see nothing but smiles and nods, it doesn't mean you've built a great team. It means the real issues are hiding in the hallway, whispered behind closed doors, or ignored until it's too late.

It's not complicated. People avoid conflict when they don't believe it's safe. That's how bad decisions get made, quarter after quarter, until there's nothing left to save.

If your leadership meetings feel comfortable, something important isn't being discussed.

I've sat in on dozens of teams tackling the same problem.

The outcome is set in the first three minutes.

Not by strategy. Not by intelligence. By something simpler.

Who asks a question before proposing a solution?

Teams that start with "What are we actually dealing with?" beat teams that start with "Here's what we should do." Every time.

The uncomfortable part? The teams that rush to solutions feel better. They look decisive. They act confident.

They just end up being wrong, more often than not.

Speed of decision is not quality of decision.

But put any team under pressure, and almost all of them confuse the two.

Mission → 1k teams to adopt weekly prod. discovery. 📈 curr: 15/1000

I use AI tools 40h/week for real work. I share what actually works.

Most AI content is either too basic ("copy this prompt!") or too technical (47-node agent workflows). I'm in the middle — practical workflows for smart people who don't want to code.

**I post about:**
• Claude Code for non-developers
• MCP integrations that extend what AI can do
• Productivity workflows I actually use

**What I'm building:**
juttu — helping SaaS companies grow revenue from existing customers.

SWE → PM → Founder. Munich.

Niko Noll

I moved my company out of Notion.

Now my docs update themselves.

Every startup I've worked at had the same problem. You write docs during onboarding, during planning, during a sprint. Then they sit there. Nobody updates them. Six months later they're wrong and everyone knows it.

Notion made this worse, not better. Beautiful docs that are just as dead.

Last week I moved everything to Obsidian. Not because the editor is better. Because the files live on disk.

Eli Mernit put it well. The architecture of an AI agent is just two things: the filesystem as state, and Claude as the orchestrator. If your docs are files, agents can work in them.

So I added agent.md files with prompts that tell Opencode/Claude Code what each doc is for and how to maintain it. The agents clean up outdated sections. Flag inconsistencies. Update references when something changes.

The docs stay alive because somthing is actually maintaining them.

We sync them over Obsidian sync.

Most documentation problems aren't writing problems. They're maintenance problems. And maintenance is exactly what agents are good at.

What's the last internal doc you actually trusted?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

I share how I use AI to build, measure, and learn faster | Founder, Product Analyst AI

Famous VC built his own CRM with AI.

He'll be back on HubSpot next month.

Michael Bloch spent 50 hours building a command center with Claude Code. Email, calendar, CRM, the whole thing.

Uses it twice a day. Saves 15 minutes.

The build was easy. Maintenance sucked. Every new feature made the system more fragile. No one maintaining it but him and a chatbot.

Sounds familiar?
Yes, I'm looking at your 34 half backed lovable projects.

Everyone's asking "will SaaS die?" Wrong question.

The real shift is supply explosion. When building costs approach zero, every vertical that had 4 competitors gets 40. Every niche that wasn't worth a venture-backed company is now viable for a lean team with AI tools.

I'm one of those new entrants. No VC funding. Small team. Building Product Analyst AI with deep domain understanding of how tech businesses use data to make more money from existing customers.

For the consumer, this is great. More specific solutions that actually solve your problem.

For the builder, the job is the same as it ever was. Solve a specific problem deeply. But the bar is higher now.

We like it that way.

Don't just ship software that theoretically helps someone get the job done. Actually get it done for them.

Code got cheap. Solving real problems didn't.

What's the one thing your current tools still don't actually solve for you?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

Anthropic just launched a legal product.

Every B2B SaaS should be paying attention.

Not because legal tech is your space. Because a model company just decided the application layer is theirs too.

Think about what that means.

The company making the AI decided it can also make the product. No wrapper needed. No middleman. They went straight to the end user.

Will they do this for analytics? For project management? For CRM? Nobody knows. But the playbook exists now.

Meanwhile, MCP is making data portable. Users can query your product from a terminal. "Death of SaaS" is trending every other week.

Some of this is overhype. Some of it is very real. And right now, honestly, I can't always tell which is which.

We're building Product Analyst AI in the middle of all this. Some mornings it feels like we're building on quicksand. Other mornings it feels like the biggest opportunity of our careers.

Both feelings are correct.

The founders who win this moment won't be the ones who predicted everything right. They'll be the ones who kept building while staying honest about what's changing.

That energy is something else right now.

Are you more excited or more terrified? Genuinely curious.

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

Last week I killed our Notion.

Here's what actually replaced it.

It starts with one file. We call it AGENT md. It's the map of our entire company. Folder structure, naming rules, key documents, and instructions for how agents should work with all of it.

The vault is one folder per area. Growth, Product, Customers, Weekly planning. Each folder has living docs that get updated continuosly and task files that get archived when done.

Three rules keep the whole thing alive:

1. Update before create. Search for existing docs first. If it exists, update it.
2. Archive don't delete. Completed work moves to _archive. Nothing disappears.
3. Link to parent. Every doc connects to its strategy doc. More connections = better context for agents.

Start of week: update This Week md with priorities. During work: create task files, link to parent docs. After completing: update the living doc, archive the task.

AI can generate endless documents. The hard part is keeping them current so the next session builds on the last one.

Full AGENT md in the comments if you want to steal it.

Are you also struggling with so much LLM output going nowhere?
This doc structure helps us, but always happy to hear suggestions on what to imrpove.

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

"Which users are about to churn?"

I asked Claude Code. It queried PostHog.

No dashboard. No funnel builder. No segment filters. I typed a question and got an answer.

Here's the setup.

One command in your terminal:
npx @posthog/wizard mcp add

It connects PostHog's MCP server to Claude Code. Log in when prompted. Done.

Next session, I started asking.

"How many users signed up last week?"
"Which features do trial users ignore?"
"Show me activation rates by cohort."

It generates the query, runs it against your data, returns the answer.

Two minutes to set up. The interface shift is real.

But that first question? "Which users are about to churn?"

PostHog MCP can't actually answer that well.

It writes a single SQL query against your event data. Counting, filtering, grouping, yes. Perfect for quick lookups.

Churn prediction isn't a query. It's pattern recognition across behaviors over time. You need a context layer on top of raw tracking data. Events and timestamps alone don't tell you who's leaving.

MCP nails the quick "I wonder if..." question. Predictions need something deeper.

Knowing the difference is the real skill.

What's the first question you'd ask your data?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

We stopped asking AI to query data.

We taught it to think like an analyst.

First version of Product Analyst AI was simple. Connect to your data, ask questions, get answers. SQL under the hood.

It worked for "how many users signed up last week." Fell apart for anything real.

"Which users are about to churn?" It would write a query. Count logins. Return a list. Technically correct. Completly useless.

A real product analyst doesn't start with the query. They start with context. They know your product's lifecycle. They know what an AHA moment looks like for your users. They know which behaviors actually matter vs which ones just look busy.

So we built that into the agent. A context layer.

It knows your ontology. Your activation signals. Your key-value moments. The stages users move through before they upgrade or leave.

Now when you ask "who's about to churn," it doesn't just count logins. It looks at the patterns that actually predict churn for your specific product.

Same AI model underneath. Completely different answers.

If you're building AI products, this is the real unlock. Domain context beats raw intelligence. Every time.
And if you can combine the two, sure even better!

What domain knowledge would make your AI tools actually useful?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

Karpathy called it a phase shift.

I'm old enough to know what he means.

Andrej Karpathy, ex-Tesla AI head and OpenAI founding team, wrote that he went from 80% manual coding to 80% agent coding in weeks. Called it the biggest change to his coding workflow in two decades.

I'm not ex-Tesla, nor part of OpenAI. But I relate to all of it.

I build Product Analyst AI mostly in English now. 
I describe architecture, a system.
So do coding skills atrophy? Is my ego as a Software Engineer hit?

It's true:
My fingers genuinly hesitate on syntax I used to type without thinking.

But here's the part I'd add.

The engineering job didn't shrink. It shifted up.

I'm more of an architect now than a coder. I design systems. I make trade-off decisions. I think about how components fit together. That was always part of the job as an engineer. Now it's the main part.

I don't write individual functions by hand for well-proven, well-documented patterns anymore. That's the real change.

It's not slop. It's a higher level of abstraction.

And the expansion is wild. I build things now I never would have attempted a year ago. Not just faster. Things that weren't worth trying before.

Is your role shifting up or does it feel like something else?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

Half of AI agents just write code.

Every other work is still being done manually.

Anthropic published their agent deployment data. Software engineering takes roughly half of all agent usage.

Marketing, ops, data analysis, sales? All under 10%.

Greg Isenberg on this chart: "every category under 10% is a department still running on manual effort and tribal knowledge. Turn one of those workflows into a repeatable AI agent. That's vertical SaaS 2.0."

Whatever you think about AI hype. Agents will be built for all of these segments. The opportunity is figuring out how.

This is what excites us at Product Analyst AI. How do you build an agent that acutally gets the job done in analytics? Not just querying data. Helping you find signals, improve your product, make better decisions.

Every product person should be asking this right now. What does a useful agent look like for your vertical? What work does it do?

Most of these segments are wide open.

What vertical would you build an agent for first?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

AI made me question everything I knew.

Then I realized that was the point.

For years I was a happy camper. Knew what SaaS looked like. Knew how B2B products worked.

Feeling confident as a product person and founder.

Then AI hits and everything is up for grabs.

How will products look? Is SaaS dead? How do you even price this?

FOMO. Fear. Anxiety. I feel all of it. Every founder and PM I talk to feels it too.

Then it clicks.

That fretting means you're conscious things are changing.

Seeing where value gets created, figuring out how software gets jobs done for users... that's our work. Was always our work.

It's just more dynamic right now.

If you're the one feeling anxious and actively learning, you're probably the one who'll help the world figure this out.

That gives me a lot of room to grow as a product builder.

I hope it does the same for you.

P.S. I'm figuring this out alongside you. You'll shape how the greatest products of the future look. Follow along if that's your jam.

Everyone's focused on AI models. I'm watching MCP.

It's the protocol that makes AI workflows real.

Every week there's a new model. Faster, smarter, bigger context window. And yeah, that matters.

But the thing that actually changed how I work? MCP. Most people have heard of it. Almost nobody actually uses it.

Model Context Protocol. It lets Claude Code talk directly to my other tools. Tella, Typefully, data sources... all connected. No API scripts. No Zapier. No glue code.

I didn't build any integrations. I just... plugged things in.

That surprised me honestly.

But here's what got me thinking as a product person.

Your product's surface area just changed. Users won't only interact through your UI anymore. They'll reach your product through Claude, through agents, through whatever MCP client they're using.

Your MCP server might matter more than your front-end soon.

The products that win next? They'll be the ones you can plug into any workflow without thinking twice.

I try to show how I actually use this every day. Real workflows, not theory. Hopefully enough to get you to try it.

Are you thinking about your product's MCP surface area yet?

P.S.
Hey, I'm Niko.
I share how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

I merged two datasets nobody combines.

It revealed which behaviors drive real LTV.

Most companies keep usage data and billing data in separate worlds.

Product team looks at feature adoption. Finance looks at revenue.
Nobody connects the two.

So you optimize for engagement metrics that have zero correlation with money.

You can often infer LTV from tracking data alone. You don't always need the billing system.

But even that, most teams skip.

Start asking better questions:

Which behavior patterns lead to highest LTV?
Which acquisition channels convert to paid fastest?
What do your best customers do in their first 7 days?

These aren't hard questions. But most teams don't have the answers.

The answers change how you build product.

We've been doing this for fast-growing startups. If you want answers to questions like these, DM me.

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

Every startup we worked with think they have a data problem. They don't. 

They have a setup problem. The data is there, it's just not structured for the questions they're actually asking.

And AI makes this harder to see, not easier:
You type a question, get a confident answer, and assume the analysis is sound. But the quality of what comes back depends entirely on what's underneath.

There are 3 levels of data questions. 
Each one needs a different setup to actually work.

Level 1: Query.
Count, filter, group. One question, one answer.
"How many trial users converted last month?" → 1,247.

Level 2: Analysis.
Compare, correlate, find patterns across behaviors.
"What do users who convert have in common that those who don't, don't?"

Level 3: Prediction.
Score users based on behavioral patterns over time.
"Which free users are most likely to upgrade this month?"

Most teams are at Level 1 and think they're at Level 2. Because the interface feels smart even when the operation underneath is simple.

The AI isn't the bottleneck. Your setup is.

We're building the layer that gets teams from Level 1 to Level 3. That's where the real insights live.

Which level are most of your data questions at?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

MCP analytics has a dangerous problem.

It feels smarter than it actually is.

The interface is incredible. Ask a question, get an answer. For ad-hoc questions, this is exactly the future.

But under the hood, every MCP call is a masked API request. It writes a query, runs it, returns the result. No memory. No state. No persistence.

Real analysis doesn't work like that.

You clean the data. Then segment it. Then test specific hypotheses against those segments. Each step builds on the last.

MCP restarts from scratch every time.

"Show me weekly active users." Great, here's the answer. "Now segment that by plan type." It doesn't build on the first query. It starts over. New query. New context.

You can work around this. Tell Claude Code to write Python, save CSVs, build locally. But now you need coding skill and analytics skill. The "just ask a question" simplicity is gone.

The chat interface for data is absolutely the future. MCP as the analytical engine behind it? Not there yet.

Know what it nails. Know where it breaks.

What's the most complex question you've tried asking your data?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

"AI > UI" was a spicy take in 2024. In 2026 it's just... what's happening.

If your product's value lives in the interface, that's worth sitting with for a minute.

Amplitude now has an MCP server. Connect it to Claude Code and just ask questions. "Which free users are closest to converting?" No dashboard. No filters. No segments. Just the answer.

Incredible for users. Less so if your product IS the dashboard.

The tension is real.

Open your data via MCP and users query it from any AI interface. Your UI becomes optional. Your edge shrinks.

Keep it closed and you're the walled garden nobody wants.

Amplitude chose to open up. Smart move. But the value now has to shift. From presentation to curation. From "here's a dashboard" to "here's what you'd never think to look for."

Dashboards won't disappear. You'll still check the same five charts every Monday.

But the ad-hoc question, the "I wonder if..." moment... these are moving to chat. And they're not coming back.

The products that thrive won't fight AI for the interface. They'll be the data layer AI can't work without.

Would you give users MCP access to your product's data, or keep the UI edge?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

Build-measure-learn is about to close itself.

OpenAI just showed us how it starts.

A small team shipped an internal product. Zero manually-written code. A million lines. ~1,500 pull requests. Three engineers. 1/10th the time.

But the code generation isn't the story.

The agents closed their own loop. Boot the app, drive the UI, reproduce a bug, record a video of the failure, fix it, open a PR, merge. While the humans slept.

Write. Test. Observe. Fix. No human in the middle.

Right now that loop is narrow. "Measure" means test results and CI passes. "Learn" means fixing what broke.

But imagine widening it.

What if agents could see usage data? Not just "does the code compile" but "are users actually clicking this button?" What if experiement results and funnel drop-offs fed directly into the agent's context?

That's build-measure-learn fully closed. We're not there yet, honestly. But the direction is clear.

This is a scary and exciting time to build. I'm choosing exciting.

What's the first piece of your product workflow you'd want an agent to close the loop on?

P.S.
Hey, I'm Niko.
I'm building Product Analyst AI and sharing how I use AI workflows and product analytics to build better products.
I also make walkthroughs and tutorials on YouTube.
Follow along if that's your jam.

taplio

The Battle

Mikhail Lvovskii

Niko Noll

Followers

Posts

Engagements

Likes

Performance ScoreMikhail Lvovskii's LinkedIn Performance

Performance ScoreNiko Noll's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Mikhail Lvovskii

Niko Noll

Followers

Posts

Engagements

Likes

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);}Mikhail Lvovskii's LinkedIn Performance

Performance ScoreNiko Noll's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreMikhail Lvovskii's LinkedIn Performance