In 2018 I brought down a production system with 400k active users. On Friday 5pm. And the senior engineers had already left the office.

What do these two stories have to do with the current AI race?

I’ll tell you.

(a) we are still early, very early and (b) shit in shit out.

Let me expand on these failures of mine. But first, the summary for those with a short attention span.

I’ve spent the last year and a half on the cutting edge of AI. I built on top of harnesses, used early coding agents, worked on memory systems and set up long running agents with complex tooling.

There is much I have learned about AI as an emerging technology but even more about human nature.

So to the first point: we are still so very early.

90% of the stuff you read online is pure FOMO. People postulating they have all these crazy AI setups that do crazy stuff. And telling you you are falling behind.

10% of people have actually unlocked incredible things with AI. It’s worth finding them and seeing what they do.

The second point: shit in shit out.

Yes, you can generate more code, more text. But just like I thought to push another release on Friday afternoon in 2018, just because you can push out more shit doesn’t mean you should. I was a pretty shit programmer back then. And it broke the platform that thousands of people wanted to use.

There is a legitimate change happening in how we write code, design, write marketing copy, etc. — but do not forget in the end this has to work for your business and for your customers. And the AI does not feel pain about bringing down your system or insulting your clients. They don’t get fired.

I am far from an AI hater. I have changed how I work, I’ve built an AI-first startup. But it is my (and all our) job to look critically at technology and figure out how to use it to our advantage. And that requires a quiet morning, a cup of coffee and some actual thinking.

In 2014 I technically understood Bitcoin. But I did not understand Bitcoin. Do you know what I mean?

It took me years and deliberate thinking time to understand the value the technology could provide.

But all I was focused on for a while was thinking how late I was. To the party. To the price appreciation, to the technology trend of digital money.

When I actually understood Bitcoin in 2019 I was certain I was the last one in the world to do so. And as history tells us, I wasn’t the last one. Not by a long shot.

LLMs are barely a couple of years old, the ones that can perform more complex tasks at least. And while we should rush to understand how to apply these tools, there is also benefit in slowing the F down.

Workflows, systems and code will evolve. But if you know what an LLM is, and maybe know what a tool call is, or how to set up an MCP... then you are definitely not early.

Anthropic released Opus 4.7 yesterday and updated their Claude app. Within 1h of use internally we encountered 3 pretty bad bugs with the software.

Pretty annoying ones at that. A longer running job (that took up a lot of tokens) just randomly stopped, for example. We paid for that inference and Claude just crashed and we had to start from scratch.

Now, they can afford that, because they are selling something literally every knowledge worker wants to buy right now. They are selling intelligence.

But still, with all their talented people they ship more and more buggy software. I’m not the only one recognizing this. We have a quality problem.

And this gets us to the core issue. AI is trained on the average. And it doesn’t get fired.

The average creative writing, the average codebase. And if it ships a buggy feature or brings down a codebase, it’s not going to get fired.

I was horrified when I brought down that platform with 400k users on it. I stayed in the office until it was fixed. I troubleshooted with a junior colleague that was still around. I wrote a post mortem and had to discuss it with the senior engineers.

Most importantly I learned from my mistakes. Turns out humans are really good at doing that.

So to summarize:

- you are not late. we are early.

- AI can help you create a lot of average quality code/words/designs.

- there is real opportunity in being early to a technology trend. But no need to stress and FOMO.

- in many places I (and probably you) am below average. AI’s average output is way better than what I could do. And it can do it 24/7.

- so quality and learning from mistakes remains something that humans (you!) should focus on. And offload the rest.

If you’re reading this, you probably subscribed a while ago when I was writing about churn. I was deep in the world of cancel flow optimization. Helping SaaS companies save users at the moment they hit “cancel.”

It worked. Personalized offers at the cancel button, dynamic based on why someone’s leaving. Some teams saw 20-30% fewer cancellations.

But here’s what kept bugging me. Every conversation I had with founders eventually landed in the same place: “We don’t actually know what’s happening in our product.” Cancel flows (while useful) were only putting a bandage on the flesh wound.

The real problem was bigger.

Teams had dashboards everywhere and still couldn’t answer basic questions about their users.

Like

• what kind of users upgrade to paid?

• what makes a user cancel? Can we predict that?

• which customers should I focus on in Customer Success?

I kept thinking, why are we only helping at the cancel button? The interesting stuff happens way earlier.

So I zoomed out! We took a hard look at where we can provide the most value right now.

What I’m building now

I’m building Product Analyst AI. It connects to your existing analytics stack, PostHog, Amplitude, Stripe, and delivers answers directly in Slack.

Not dashboards. Answers.

You ask a question in plain English. “Why are trial conversions dropping?” “Which features do retained users actually use?” “Who’s about to churn?” And you get a real answer, reviewed by a real analyst, in your Slack channel.

We’re running pilots right now. Small team, three of us. Still very much figuring things out. But the goal is the same one I’ve always had: help companies make more money with the users they already have.

Cancel flow optimization was one narrow slice of that pie.
Now we are going after a bigger problem. And it’s so nice, because early clients just CARE more, cause it is more important to them.

What this newsletter will be

I’m restarting this newsletter. Here’s what you’ll get:

How I actually use AI in my work. I spend 40+ hours a week working with Claude Code, AI agents, and MCPs. Not in a “look at this cool demo” way. In a “this is how I do my actual job” way. I’ll share the workflows, the prompts, the tools. Stuff you can try yourself.

What I’m learning from building. The founder journey, honest version. Pilots, pricing, the things that work and the things that don’t. Real numbers when I have them.

Patterns from the work. Anonymized insights from the analytics work I’m doing with teams. What product leaders are missing in their data. Things I keep seeing across different companies.

Short takes on product analytics. Why dashboards don’t work. What “data-driven” actually looks like in practice. Opinions, but grounded in real experience.

Posts will be short. I’ll try to post weekly. If I don’t have anything useful to say, I’ll skip a week rather than waste your time.

If this isn’t your thing anymore

Totally fine. The unsubscribe button is right there. No hard feelings.

But if you’re a founder, product person, or someone curious about how AI is changing the way we build and measure products... I think you’ll like what’s coming.

The mission hasn’t changed. Helping companies grow with the users they already have. The approach just got a lot bigger.

Thanks for being here.

Onwards,

Niko

And honestly? That’s great news for us.

Every SaaS should have a cancel flow. One that actually tries to retain revenue and figures out why customers leave. Not some afterthought modal with “Are you sure?” - a real conversation at a critical moment.

We’ve decided to make this our main thing with Juttu.co . Cancel flows. That’s it.

We have users. They’re onboarded. Revenue is coming in.

So what’s the problem?

The Awareness Levels I Didn’t Know About

Here’s something I learned the hard way.

Finding people with a churn problem isn’t the bottleneck. Most SaaS founders know their retention could be better. That’s not the issue.

The issue is awareness levels.

I didn’t know this was a framework until I ran face-first into it. There are layers:

1. Problem-aware — They know churn is an issue

2. Solution-aware — They know a cancel flow could help

3. Product-aware — They know about us, specifically

We assumed most people were solution-aware. They’re not.

A lot of founders know they lose customers. Fewer know that a well-designed cancel flow could save 10-20% of those cancellations. Even fewer know that tools like ours exist.

This changes everything about how we sell (get customers).

Why Outbound Doesn’t Work (For Us)

We done the outbound motion so far.

LinkedIn DMs. Qualified everyone thouroughly. Scan their product page etc. - The whole playbook.

Here’s what happened: You get on a call with someone who has a churn problem, but you spend the entire call educating them on the solution. By the time they understand what a cancel flow even is, the call is over.

You can’t educate your way through a sales call when the prospect isn’t even solution-aware yet.

At least we can’t as self-funded startup.

This is why we’re doubling down on content. SEO. On being visible. On building in public. We need to find people who are solution aware or create solution awareness before the sales conversation even starts.

The Integration Problem

Now here’s the part I underestimated.

For us, integration is activation.

We have a code-based SDK. You connect to your payment provider. You add our widget to your cancel flow. It’s not drag-and-drop.

And even for people who are committed — who’ve signed up, who are excited, who want this to work — actually getting it live takes intentionality.

This is the thing that people don’t talk about when they say everything is vibe-coded now.

Sure, you can build fast. But you still need to test. You still need a running product with real customers paying for it. And even something that seems “unimportant” like a cancel flow still takes real effort to integrate properly.

I thought our sticky revenue model would be the greatest thing ever. Lock them in, they never leave, compounding MRR forever.

But stickiness has a price: **activation is hard**.

The same thing that makes customers stay makes them harder to activate in the first place.

What I’m Taking Away

If you’re building something:

1. Know your prospect’s awareness level. Problem-aware, solution-aware, or product-aware? Each requires a completely different approach. You can’t outbound your way past a solution-awareness gap.

2. Integration difficulty = activation difficulty. If your product requires technical setup, budget for that. It’s a bigger hurdle than you think, even for motivated users.

3. Sticky revenue is a trade-off. Yes, it’s compounding. Yes, it’s beautiful. But the integration barrier that creates stickiness also slows down your path to activation. Pick your hard problem.

We’re still acquiring customers. Still growing. But I’m recalibrating my expectations on velocity.

Some products are slow to activate by design. That’s not a bug. It’s a trade-off you make when you choose to build something that actually integrates deeply.

What’s Next

We’re all in on making customers more money with existing users.

That is the real problem we are after. So, you might see us coming out with other solutions for the same ICP :)

Cancel flow are one small element of that.

For cancel flows specifically: Creating more content around the solution (SEO). Making people aware that this category exists. And improving our onboarding to reduce the integration friction as much as possible.

If you’re building a SaaS and you don’t have a cancel flow yet — you’re leaving money on the table. And we’d love to help.

Subscribe if you want to see how this bet plays out.

Onwards,

Niko

The internet is getting noisy. Social media is getting noisy. We are trying to use AI to actually enhance our authenticity and reach out to find clients. We do so by using tools like Teller to do screen recordings and using Wispr Flow for dictation.

Here is where we are right now, what we're trying, and what's working.

where the company is right now

It's been a while. I've been busy building and trying to acquire customers. The pitch for Jutu.co has been working increasingly well since we are actually focusing on trying to save customers money.

Spoiler alert: that works better than trying to sell a vitamin.

Funnily enough, the biggest challenge now is actually to get people's attention because cancel flows in particular are being seen as something you can build in-house or a self-problem. We do a lot of work to build the best cancel flow experience out there. But people are looking for AI features and are looking for revolutionary agents, and I think right now the scope of expectation and B2B SaaS is kind of overblown right now.

how we are finding customers

after throwing a bunch of spaghetti against the wall (different channels, formats, etc.) this is the current strategy for juttu.co

→ Glad for any challenge of the approach, or feedback overall

Results so far: Getting some calls booked. Nothing like reliable pipeline yet.

Content & Reach Out

• run dripify automation to add people that are followers of Kyle Poyar

  • ask specific qualifying question right up front

    • ```Hey , is churn a topic for you right now?

      We're a small team working on an AI-cancel-flow that boost NRR and finds out why people leave.

      Just looking for feedback on the product right now.

      Cheers from Paris```

    • 1 more message focused on retention outcome we are seeing with existing customers

    • follow up 2 months later: been a while, how is retention at %%company%% going.

• post LinkedIn posts 3-7 times a week around pillars

  • pricing,

  • b2b saas,

  • retention

  • meta theme: AI and how it affects 3 other pillars

  • 4th pillar for the soul: founder journey posts

  • notes so far: visualizations and video as focus. exaclidraw is like the simplest ever. also drawing simple stuff works better than trying to be cute. basic concepts > complex ones. feels like maybe people scroll by with half an eye and only have 1 sec to absorb it. so the simpler it is to grasp the greater the engagement.

• sprinkle in specific post about cancel-flows as retention technique once a week

  • DM people who engage with ask: “want to give feedback on the tool”

  • (that’s the playbook a guest on a TGS webinar was giving)

• ah, an manually adding and DMing people

  • in French and German is working quite well. same message, not personalized. but localized.

  • Keeping the tone of voice super authentic and real. I'm often even adding video recordings or screen recordings with Tella to make sure that people do not mistake us with AI bots.

Keeping it authentic

The challenge that I am taking on right now is to be as real and human as possible, leveraging AI to scale that authenticity. This post, for example, was all done by dictating Wispr Flow. That is a way to use AI to just get your thoughts out in a more raw way, without having AI be the arbiter of your tone of voice and the facts that you actually share.

I hope you're doing super well finding customers and keeping it authentic. It’s the only way!

I will for sure.

Niko

I’m writing this for fellow founders/builders/PMs who are juggling too many opportunities and wondering when to pivot versus when to niche down. If any of this sounds familiar, read on I guess.

Snapshot of the Road So Far

• v0 – “Continuous Product Discovery”

  • Built a B2B Product Discovery tool.

  • Landed paying customers, but their teams didn’t feel enough pain to champion it internally.

  • Lesson: first traction ≠ deep problem fit. Look for urgency not curiosity. Especially if you are working on an AI product.

• v1 – AI Surveys (first Juttu pivot)

  • One LinkedIn post → hundreds of trial sign-ups, constant word-of-mouth inbound.

  • Problem: Our ideal users (PMs & UXR folks) want quantitative reports/CSV files, while Juttu generated qualitative insights.

  • Competitors like ListenLabs doubled down on the qualitative part (interviews) that path didn’t feel natural to us.

• v2 – AI-Personalized Cancel-Flows (today)

  • Same underlying tech, but super-focused on subscription churn.

  • Early calls show a sharp pain: “how do we keep revenue we’ve already earned?”, “We should do something about churn!”, “So far we dealt with cancellations manually but it’s a lot of work”

(here is a screenshot of what users get to see once they’ve implemented the cancel flow. just so you can imagine what we’re talking about)

But now towards what i’ve learned.

3 Lessons I Keep Re-Learning

1. Pivoting ≠ Flailing

   • A pivot is only a “dead end” if you stay half-committed.

   • Each version sharpened my understanding of who feels the pain deepest.

2. Choose a Problem, Not a Solution

   • Starting with “AI surveys” was sexy tech; starting with “stop churn” is a survival-level problem. As in people care

   • The tech remains, but the framing changed everything—especially in sales calls.

3. Go All-In (Until You Change Your Mind)

   • I’m publicly betting on the European Churnkey alternative.

   • If data disproves this thesis, I’ll change direction loudly and quickly. Silence helps no one.

How I’m Executing the Current Bet

• Niche definition

  • SaaS & subscription companies with 3–30 % monthly churn.

  • Self-service plan available (pricing < 500$ a month)

  • Starting with EU based companies

• Proof plan

  1. Ship a live cancel-flow demo inside two weeks.

  2. Interview ten CRO/Product Managers or CX team leads focused on retention

  3. Convert three to pilot customers within 30 days.

What’s Next

• Build out the adaptive flow logic.

• Publish the first teardown of an anonymized customer’s churn data.

• Share every “no” I get along the way - because the rejections teach faster than the wins.

Ask Me Anything

If you want to nerd out about stuff like this → My DMs are open.

If you know someone who is looking into reducing subscription churn … lmk! Would love to see if what we’ve built is a fit.

Niko

pivot fast but also commit.

I guess that’s the interesting duality I’m learning about …

go watch it, i think you might like it!

But here’s the real story behind the scenes: this year already feels like it’s been 10 months long.

Between restructuring the company, pivoting away from our old product, reducing the team, and rebuilding runway…

i’ve had moments where i was this close to quitting.

not because i don’t believe in juttu.

but because getting from 0 → 1 again after killing something you poured your heart into?

it’s brutal.

there’s no manual for this.

no step-by-step guide that tells you how to rebuild momentum from scratch.

i needed to push hard.

not because it was wise. but because standing still would’ve killed the spark.

i’m proud of that push.

not because it worked.

but because it reminded me that i’m not helpless—and maybe that’s all we need to remember sometimes.

if you’re in the middle of a reset yourself,

know that it’s okay if it doesn’t feel strategic or smart or well-paced.

sometimes you just need to move.

Keep shipping

Niko

Subscribe now

Build in public update

Juttu is evolving - and so is our thinking.
I’ve been heads-down building and honestly… not great at keeping you all in the loop.

Here’s where we are:

🔷 People are using Juttu every day
🔶 We haven’t nailed the sticky use case
♦️ still at 200$ revenue

(see what I did there with the colors?)

Public Changelog

You can always see what we are up to, what we are shipping (and some of the planned things) on the changelog site.

Ai looking for a problem

We’ve been talking too much about what the AI can do - and not enough about the problem it actually solves.

Obvious, right?

Hard to not focus on shiny features that were science fiction two years ago.

What we have shipped:

• Smarter, more context-aware survey questions

• Full internationalization: 20+ languages supported

• Early churn analysis (analyzing satisfaction survey)

Next: more talking to users.

We’re looking for the places where hyper-personalized forms actually help with a CRITICAL workflow - by saving customers time, money, or mental energy.

This part’s hard. But it’s the good kind of hard.

We are very much in the desert of the startup building phase.

Thanks for following the journey. I’m excited to keep building in public.

Have you tried the product? It's free for the first survey (for now).

Your feedback is very much taken into account here, so shot me a DM or comment below.

Let's make this a build-in-public thing.

PS: let me know what you are working on, and if I can help!

Launching felt great. And while I knew that launching comes in waves and is a continuous effort, I underestimated the energy it takes to keep momentum up.

I think this is what a lot of founders chose to not talk about.

You get first signals from the market, excitement even, but no willingness to pay or really switch to your product.

This is where we are. We need a strategy.

Recharge

Riding the wave of users signing up to juttu.co I was so excited to iterate super fast. I identified a few use cases just by looking at what kind of surveys people were creating.

But then my nervous system started acting up.

My eyes got tired, my neck cramped up and I could not focus anymore.

Product Strategy is just taking bets

After this unwanted slow-down I feel recharged. While it was not planned to take it “easier” for almost 2 weeks, I know can see how much perspective I got from that step back.

This week we adjusted the product to fit a particular use case. Instead of a generic form tool, we want to get this to solve a specific problem.

It’s early days, but first conversations with the target customers seem promising.

I’ll report back on the results.

But what I can already say: Opinions are cheap. Many people have them. Also about our Product. Especially on LinkedIn. Product Strategy is not about having onions but about making hard decisions about what not to do.

There is so much opportunity, so much hype in AI, so much seemingly promising feedback.

But in the end the question we ask ourselves is:

What is the one thing we need to do to solve a problem that people want to pay for.

And everything else is deprioritized.

Behind the scenes #3 (BTS3) is out

In this edition of the video logs, I walk you through I’m thinking about Product Strategy right now, first principles and the AI-forms market in particular.

Go check it and let me know in the comments what you thought. I want to get better at these videos and could really use your feedback!

Love

Niko

The MVP is all the rage but first impression matter

Just a few weeks ago I had to make the call to not launch juttu.co on time.

Instead I delayed the launch to improve quality.

Here’s the story of why and how I think about Quality vs. Speed:

Every weekday, every weekend - coffee, coding, lunch run, coding, walk to winddown, meditation, sleep.

I’d collapse at night exhausted and have a few hours of super deep sleep.

Then I’d wake up in the middle of the night (usually exactly at 4am), excited about a new feature or something I still needed to build.

It was an intense, deep-focus period. Probably the most productive I’ve ever been.

And now? The lull.

Post-launch is different. The rush and the excitement of building fades, and what’s left is ambiguity.

Launch day is never as big as you imagine.

We had solid traction on LinkedIn, 10k impressions, 400 website visits of the launch post, 160+ surveys created, 20+ signups.

But PMF doesn’t just happen overnight.

People are loving the product, say it’s fun for the *first time to create surveys* and are slowly discovering the value and the use cases for the “conversational” context-aware follow-up questions.

We even had our first users “hacking” the tools to serve their needs in totally unexpected ways.

All that is exciting, but never the less, we have low 2 digit accounts and 0$ MRR at the time of writing.

What was once clear is now uncertain.

During the build, every next step was obvious. I had a very clear vision. And to be honest wanted to prove to myself that I could build something impressive.

And instead of being hesitant, like before our pivot, I wanted to really make a statement, with a lovable v1.

Ok, big effort, I’ve shown myself I can still code and move fast.

Now, it’s about figuring out the next move in the market.

The shift from execution to iteration.

Instead of just shipping features, it’s about getting feedback, refining positioning, and pushing adoption.

Especially the “growth” part is something, well, I’m still learning.

This phase feels different—but it’s part of the journey.

To be honest I feel depleted. I’ve been taking some half days to explore the city (Seoul, South Korea) and hike in nature. It’s been so needed.

Building in public means sharing not just the highs of launching, but the lows of what comes next.

PS: Go take a minute and enjoy the ride. That’s what I’m doing. The lull is a place to recharge and go touch grass.

Not sure if you know Daniel Dalen. He runs Ecomflow, an e-commerce fulfillment company, and pulls a ridiculous amount of business through POV-style vlogs on YouTube. No sales pitches, no over-produced nonsense—just a camera, his work, and insights on running a fulfillment business.

It works.

Felt natural to me.

So I’m documenting the daily life of building this startup on YouTube.

In an age of AI generated generic stuff, I think real personality and building real trust is all that matters. Long term at least.

my first video

LinkedIn vs. YouTube: Two Different Beasts

I’ll keep LinkedIn going. It’s predictable. Works well for B2B founder-led distribution. I know how to play that game.

But YouTube? Different animal.

• Hard to track impact

• Hard to scale (one video at a time, no “comment to get playbook growth hacks”).

• ROI takes forever

But it compounds.

And it builds something different: proof of work, credibility, and inbound talent.

What’s the Actual Goal?

Do not expect customers from this short or medium term.

More about:

• Hiring talent (people work with people they trust)

• Compounding (youtube is evergreen, ads or LinkedIn isn’t)

• Trust through doing, not telling

This is an experiment. Weekly uploads. No fluff. Will share learnings over the next year.

If anyone’s done this before—hit me up. Would love to learn from people who’ve actually grown a channel this way.

Let’s see where this goes.

Hope you have a great weekend.

Cheers from Vietnam

Niko

On a new venture, a new SaaS. It’s always scary and exciting and makes things real.

I try to launch quickly, to rip the badaid off, but at the same time I know: First impressions matter.

So yesterday I released a lot of UX improvements that mostly focused on

• making the ONE thing very obvious

  • the one thing I want people to remember. Peoples attention span is not as high as you might think. So, it’s worth just making sure you hone in on one thing. For juttu.co that was “Juttu asks real-time generated questions to provide deeper insights.”

• nothing breaks on mobile

  • even in B2B SaaS, people check stuff first on the phone. They send it to their boss, who checks their mails on the phone. So, you gotta be mobile friendly, even in B2B.

So, how did the launch go?

Great. We didn’t go viral that’s for sure.

Mostly posted on LinkedIn. I have 7k+ followers there and really the only channel where I can reach a lot of people quickly.

We didn’t go viral, that’s for sure. But we got some eyeballs.

Here are the stats 12h post launch:

Here are the website stats 12 hours post-launch.

Conversions are estimated a bit low here, as many people try out the survey creation straight on the landing page, which is not included here.

We had around 30 surveys created in total.

Why this is good news

Nothing broke.

Checking the logs and monitoring feedback channels, nothing went wrong. No survey creation failed, no obvious UI issues, sign-up issues etc.

People got the point

I got a dozen of DMs pointing out how cool the “real time follow-up questions” are. Which, if you remember, is the ONE thing I wanted people to walk away with.

Stay top of mind

People do not create and publish surveys every day.

Now I’m on top of some people’s mind. And it is my job to stay there. So when they create their next survey, they’ll give juttu.co a shot.

Launching as a sequence of events

I was told by somebody smart that launching ONCE is a fallacy. Launching is a chain, a sequence of launches.

This one, the first one, was to get the product out there, visible, in the hands of real people (who are not my friends I forced to test it).

And to see if something breaks.

Now it’s time to improve the product, to overcome people’s objections to adopting the tool. And I’ve gotten a few objections already!

And then also just building brand, continuously getting eyeballs on the product.

After the launch is before the launch.

Onwards.

I’ll keep you posted.

Niko

PS: Obviously check out juttu.co and let me know how I can make it more useful to you!

It’s all about:

• Launching (several times)

• Lowering the barrier to get started (in 3 ways)

Follow along as we launch and grow juttu.co !

Subscribe now

You lucky person can use for free (and w/o an account) at juttu.co

But what I wanted to share, what I think is something most people don’t talk about is: Fake hype and AI agents.

In short: What is an AI agent

It’s a systematic way to use LLMs (or other ai models) and normal code together. The intention is to produce a system that combines the benefits of code (it always does what its told) and the benefits of AI (it creates and understand ambiguous stuff).

Why everything is kinda an AI agent

In every product you use, if there is AI involved the developers working on it will have a major issue. It’s that the AI does unpredictable stuff, and that the user does unpredictable stuff.

Two parties that doe unpredictable stuff is too much to handle for a code base. Too many edge cases.

But imagine now doing several AI steps after each other, where the second step depends on user input + the output of the first AI operation.

Then it gets really wild and unpredictable.

So we need to control the AI. And as soon as we attempt to do that, we are already building an “AI system” and not just a prompt. Looks a little like this:

So, the complexity differs, but wherever you see AI embedded, it’s probably some form of AI system like this - which people call agents.

Now, there are purists that somehow define agents as where the AI autonomously does everything without any human intervention. In which case, AI agents do not exist yet, so it’s kind of useless to talk about them.

Designers, Product People you need to study this

Now, while there is a ton of hype, building ai systems is going to be part of almost any product in the coming years. Not cause of the hype, but because it’s useful.

Just try out juttu.co, which is still early on, and tell me that you’ll not want survey analysis done by ai first. Sure, you can download the csv .. but why not have AI summarize the findings first?

And building my first 3 AI systems I’ve learned a lot.

I built

• a survey generation system with 5 steps and multiple branches.

• a conversational survey system that has to work “on the fly” as the user responds to user questions but needs to be context aware

• a survey analysis system that needs to extract trends, insights from complex data.

And these learning are not just about the code, they influence the UI/UX. How does it flow? How do we control user expectation, user input? How do we balance speed and quality (important!).

So I can only say: Go learn these things, play with these interfaces! They will be everywhere. Look through the hype.

This Anthropic post is a great read on it, that’s very understandable.

And shoot me a message if you want to chat about this :)

niko@juttu.co

Hope you are as excited as I am about how much better products this is going to enable!

Happy building,

Niko

Put that out there about 1.5 years ago. Guess what? Never actually explained how I planned to get there and why. So let's do that in brief here today.

What was the plan?

The original roadmap in my head was something like:

• Hands-on coaching. Start by doing all the unscalable stuff, like working one-on-one with teams, rolling up my sleeves, and actually solving problems up close.

• Scale my brain. Take what worked and turn it into something that reaches a bunch of teams at once—content, a course, something like that.

• Build SaaS. Eventually, package that know-how into software that supports teams so they can do the weekly experiments on their own.

Now fast-forward: After 1.5 years, where am I?

Reality check

Didn’t set any goals with timelines attached. so am I behind my goal? I don’t know. But 1000 feels far away that’s for sure. I guess I expected to be further along. To have made more of an impact by now.

• I did the hands-on part. I’ve coached 19 teams directly. That’s a lot of Zoom calls and Slack threads.
• I tried the SaaS step early. We built a product, got 5 teams to try it, and guess what—nobody gave a shit. Well, not enough people. Want the messy details? Here’s the post-mortem (link). Knock yourself out.
• About the course thing: I haven’t done it yet. Why? Because everyone and their mum is selling a “how to be a better PM” course, and I’m not convinced anyone needs yet another one. Maybe I’ll do it, maybe I won’t. People are tired of endless “masterclasses”. So consider that a tbd.

So that leaves me with what I love the most: software. I’m a product geek at heart and I a personal goal of mine is to build software companies for the coming decades. Even though Juttu’s first iteration didn’t catch on, I’m not done tinkering. I’m already playing with new ideas.

Still the right mission?

Why does this mission even matter?

Because most organizations waste their time on bullshit that doesn’t matter if they’re not building what people actually want. Period. All these fancy frameworks, certifications, and “look at our process!” hype mean shit if the end product sucks. Helping teams focus on outcomes and run actual experiments is what makes a difference. It makes them build stuff users want (to pay for). Less waste, more real value. And yes, I want to capture a small part of that value. I believe I can turn this into a sustainable business. Win-win.

Timeline reality check: I set that 1,000 team goal without a real deadline. Big mistake. Now, 1.5 years in, I’m at 24 teams. That’s fine, but I need more accountability.

So here’s a new milestone: by the end of 2025, I want to be at 200 teams.

Takeaways

A few takeaways from this entrepreneurial dance I've been dancing the last 3 years:

• You need a north star: Call it a mission, a big hairy audacious goal, whatever. Without one, you drift. I know I do.

• Set deadlines: Just saying “I want X” without a timeline is an invitation to wander aimlessly. Guilty as charged.

• Move faster: Without clear milestones, you lack an invalidation event. No clear point when you have to look in the mirror and say: "We're not on track. Pivot or persevere?"

• If your market doesn’t care, you’re not a hero: If people don’t want what you’re building, no amount of personal conviction or LinkedIn hustle will save you. Trust me, I tried. Again, check the post-mortem.

So that’s where I’m at. I’m still on this 1,000 team mission. At least until the end of 2025.

I still believe in it.

I’m just getting a bit more ruthless with timelines, product experiments, and figuring out what the hell people actually want.

Let’s see where I land by the end of next year. I'll document it. So if that sounds interesting go subscribe to this here newsletter.

Subscribe now

But if you are here for business, ignoring that stuff really can pay off. Here is how it did for me.

After 3 years as a Product Coach / Consultant here my tactical lessons on getting clients.

Man does it feel good writing again.

Getting into the habit of producing content is a goal for many. Building a personal brand is what many recommend. And it’s hard.

But sometimes is important to shut the fuck up as well, if you don’t have anything to say. I didn’t have anything to say for last almost 2 months.

Why you ask? Had to shut down a SaaS product close to my heart. Had to process, learn from it, take the hit and get back up. Read the post-mortem here:

So, after not posting for the longest I have in 2 years, how about a post on our favorite blue platform of cringe, LinkedIn.

I’ll tell you how I made 61512 Euro from LinkedIn in the last 2 years, my posting schedule and my content strategy. But first some data.

My “personal branding” in 2024

High-Level Metrics

• Total Impressions: 794,739 views across your posts over the year.

• Total Engagements: 7,461 interactions (likes, comments, shares, etc.).

• Average Impressions per Day: Approximately 2,189.

• Average Engagements per Day: Approximately 21.

• Followers: 6796 (started with 1k in ‘23, 0 followers in ‘22)

• Inbound Leads: 6 over the year (2 converted to paid)

But do the numbers really matter?

Eh, kinda. WHO you are reaching is more important though, I agree. Let’s look at that:

So, I’m reaching product people, good!

This is what I’m proudest off. A high seniority level. Lots of Product Leaders. I don’t do posts about the 101 of Product Management. I don’t care about it and so does my audience.

How I get clients off LinkedIn

So in 2024 I converted 2 clients off LinkedIn. And 2 in 2023. Together that came out to be 61512 Euro of revenue. There are more clients that I cannot directly attribute to LinkedIn but that might have been influenced by my posts.

Nobody is just going to wire you money

Consulting and coaching gigs are big deal sizes. The smallest deal I’ve done has been 10k. So, no, nobody just buys your services off a single selfie-post on LinkedIn. At least not my selfies.

It’s all about trust.

As a solo-coach it’s all about building authority in a specific field and building trust.

If I’m considering getting a product coach there are many questions floating in my mind. But two important ones are:

• Do I know you?

• Do you know what you are talking about?

I’m trying to make it easy for potential clients to answer these questions. With a resounding YES!

Do I know you?

This is more about “networking” than content publishing.

Through LinkedIn I’ve had hundreds of casual 1-1 messages with PMs and Product Leaders. They ask a simple question via DM. I try to be useful. 99% of them never turn into business. That’s ok. They now “know me”. At least a bit.

Through LinkedIn I’ve done online events, podcasts and in-person meetups. I’ve shook many hands. 99% of them forget about me. That’s okay.

I’ve been invited to speak at conferences, but so far, I declined. (controversial opinion redacted)

You get the point. This is a long term game. Being honest, authentic and showing up day after day builds up the surface level of those encounters that lead to actual deals.

Do you know what you are talking about?

This is where the content part of the strategy shines.

Continuous showing up and sharing analysis, opinions and resources on Product Discovery shortens that conversation.

I’ve never once had a conversation that included asking me “so, what is your skill set as a coach?”. I clearly can’t shut up about Product Discovery. It’s pretty clear I’m more versed in that niche than 99% of people.

The actual Strategy

Here’s is basically what I’ve been doing for the last 2 years.

My results are not like Leah Tharin’s or others with 100k followers that charge 10k an hour. I think they are relatable and reproducible though.

LinkedIn Daily Routine

Post

• Take a post that was already done (in buffer) and publish it or schedule a new one

  • 3-5 posts per week

  • varied formats (2 photo, 1 text, 1 video, 1 carousel)

• Go to https://www.linkedin.com/in/niko-noll/recent-activity/all/ and reply to comments on latest posts

• Send connection request to everyone who commented

Engagement

• 10-15 comments a day, by going through Creators to engage with (this is one of many lists)

• 5 comments on people comments (on big creators accounts)

• send connection request to the person who made the comment

• check DMs, reply for max. 15min

That’s roughly 1h a day if I’m quick. Talking to many other coaches with 30k-50k followers, they do roughly the same. So it’s what it takes, I’d say.

If you’ve made it this far, you should definitely follow me on LinkedIn.

What to post about (content strategy)

In my experience, it’s mostly just doing it. Finding out what you enjoy writing about, what people engage with (and sometimes those are not the same things).

But a content matrix helps. Having …

• a mantra

• 4 content pillars

• with 9 sub-pillars each

• and clear content format ideas for each

… that can all be done in a google sheet. I found and adapted this matrix online.

Here, you can have mine.

See you on LinkedIn in 2025?

That’s everything valuable I can think off. So I’ll shut up now.

7-10h/week is a lot of time for “social media”. But over half my revenue has come directly from LinkedIn. Worth it? Sure.

And while this wasn’t part of this post - I met some incredible people along the way. Online and in real life.

So in case you were waiting to start (or re-start) your LinkedIn journey, hope to see you on LinkedIn in 2025.

Ah, and subscribe to this newsletter if you liked it. I’m just getting started. Would mean the world.

Subscribe now

1. Overview

Juttu wasn’t a rocket ship. It was a self-funded project. That meant we had to work on the side to extend our runway, which slowed us down. Still, by May 2023, we were interviewing product managers and finding real pain: teams weren’t talking to users enough. By June, we felt confident we had problem validation. We quit our jobs and committed to building an MVP.

By April 2024, we had our first paying client. By May, we had our second. It felt like progress. But by September, we had only two paying clients—far from the 10 we needed to keep going. Conversations with those clients confirmed our suspicion: we hadn’t solved the problem well enough to scale. We decided to shut it down in November 2024.

2. What Went Right

The Product Worked

For the right people, Juttu did what it promised. Our two paying clients loved it and used it regularly. That felt good.

We Made It Happen

Building a fully remote SaaS company from scratch, across three countries, was no joke. Incorporating the company? Figuring out international tax laws? Navigating time zones? Harder than expected, but we made it work.

The Team

We were scrappy and resilient. Founding (and I guess first time founding) is such an emotional journey. Your self-worth (at least mine) is very much wrapped up in the products success. Shutting down was tough, but we stayed kind towards each other and made the decision together.

3. What Went Wrong

Product-Market Fit

Rob Snyder’s PMF levels nail it:

Level 1: Don’t have a customer case study that’s worth replicating.

Level 2: Have a customer case study but can’t consistently replicate it.

Level 3: Can replicate a customer case study, but it’s not a “hell yes” every time.

Level 4: Can replicate a “hell yes” customer case study but need a growth lever.

Level 5: “Hell yes” case study + growth lever, holding on for dear life.

We never made it past Level 2. Our two clients loved the product, but we couldn’t replicate their enthusiasm with anyone else. No PMF, no scale.

Calendly integration

We wanted to be clever and not rebuilt the whole scheduling part of the product. But start as a Calendly integration. After building an API integration for months, we got shut down by a change in Calendly’s open api usage terms of service. Building our own scheduling and login increased the scope of our initial product.

Sales, Sales, Sales

We didn’t realize early enough that this was a sales-heavy product. Juttu wasn’t plug-and-play; it had to integrate deeply into workflows. That meant long conversations, customizations, and hand-holding. We weren’t ready for that level of effort.

Pricing

To make the numbers work, we needed to charge $200 per product team. That’s $1,000 a month for a typical product org. The problem? For most teams, talking to users wasn’t painful enough to justify that cost. They liked the idea but didn’t need it.

Market Dynamics

We had three more clients almost ready to sign. After weeks of demos and negotiations, they dropped out—internal restructuring, layoffs, budget freezes. Things outside our control, but a reminder of how precarious the market was.

4. Lessons Learned

Separate Problem Validation from Solution Validation

Teams not talking to users? Real problem. Automating interviews? Not the solution. We validated the problem but jumped to a solution that didn’t stick. Internal problems (such as access to users) and a lack of accountability for teams to actually do Discovery were the real problems.

Winning Isn’t Binary

PMF isn’t “you have it or you don’t.” It’s a spectrum. Some of our competitors also had slow growth and challenges, but maybe just enough to keep going. The decision to stop is harder when you’re not failing outright—just not growing fast enough.

Don’t Build a Vitamin

We built a nice-to-have (a vitamin) instead of a must-have (a painkiller). Teams liked the idea of Juttu, but when it came time to buy or put in work to make it fit their workflows, they hesitated.

Market Maturity Matters

The market wasn’t ready. Product discovery tooling is still immature. Some teams care deeply about talking to users, but the broader market hasn’t fully grasped the value yet. Until they do, adoption will be slow.

You Have to Live It

A lot of these lessons sound obvious in hindsight. But no blog post, book, or mentor could’ve taught us what we learned by doing. Founding is personal. You have to go through it yourself.

5. Gratitude and Closure

The past 1.5 years have been a crash course in starting and running a company. Founding isn’t just about building a product—it’s about working with co-founders, navigating tax laws, staying motivated, and dealing with failure. It’s about all the messy, human stuff.

I’m grateful to my co-founders and our clients for coming along for the ride. Shutting down stings, but it left me with one clear thought: this is just the beginning.

I asked myself, how would I regret starting Juttu? The answer was simple: by stopping the entrepreneurial journey. If I keep going and apply these lessons, this experience will have been worth it. The mistakes we made? I’ll avoid those next time. And I’ll make new ones. That’s the game.

Building something people want—enough to pay for—is hard. It’s also the best job in the world. Onwards.

Subscribe now