DEV Community: Markus Kellermann

The Thing You're Avoiding Is Probably the Thing That Matters Most

Markus Kellermann — Sat, 31 Jan 2026 22:43:58 +0000

I spent 2 months optimizing my startup's SEO.

Landing pages. Content. Keywords. All the comfortable stuff.

Then I ran an SEO audit:

On-page SEO: A+
Backlinks: F

I'd been avoiding outreach because it felt awkward. Finding sites. Writing personalized emails. Getting ignored. Following up. Getting ignored again.

But I forced myself to do it anyway.

Results after 30 days:

Domain rating: 7 → 23
+415 quality backlinks

The Uncomfortable Truth

Google doesn't just rank good content. It ranks content other sites vouch for.

You can have a perfect site and still lose to someone with better backlinks.

The thing I avoided was the thing that actually moved the needle.

Why This Matters More Now

Ryan Dahl (creator of Node.js) said it last week: "The era of humans writing code is over."

If anyone can build a product with AI, what's actually left as a moat?

Distribution.

Your SEO (or GEO). Your audience. Your brand. Your ability to get attention in a world full of noise.

The next decade won't be won by the best engineers. It'll be won by the best marketers who can also ship.

The Pattern

Most founders do this. We optimize the comfortable stuff and ignore the uncomfortable stuff.

For me it was backlinks. For others it's cold outreach, posting publicly, or talking to users.

The pattern is the same: the thing you're avoiding is usually the thing that moves the needle.

Building Convo (AI meeting assistant) and MyPicNow (AI headshots) from Barcelona. Zero ad spend, all organic.

Why I Built MyPicNow (And You Might Want to Try It)

Markus Kellermann — Sat, 24 Jan 2026 22:55:20 +0000

The Moment I Knew I Needed a Change

I was applying to speak at a tech conference when I realized my headshot was embarrassingly bad.

You know the one — taken 4 years ago, weird lighting, slightly blurry. The kind of photo that screams "I don't take my online presence seriously."

I needed a professional headshot. But here's the thing:

Photographers: $200-500, plus scheduling hassle
DIY selfie: Still looks unprofessional
Existing AI tools: Results looked like plastic mannequins

None of these options worked for me. So I built my own.

Enter MyPicNow

What started as a weekend project turned into MyPicNow — an AI headshot generator that actually produces realistic results.

The concept is simple:

Upload 10-15 selfies
AI trains on your face (~15 minutes)
Generate unlimited professional headshots in 60+ styles

The result? Photos that look like you hired a photographer — not like an AI fever dream.

What Makes It Different

After testing dozens of AI headshot tools (yes, including the competitors), here's what I focused on:

Realistic skin and lighting — Most AI tools make you look like a video game character. I obsessed over making output indistinguishable from real photography.

Fast turnaround — Your AI model trains in ~15 minutes. Individual photos generate in 8-10 seconds.

Actually looks like you — This sounds obvious, but you'd be surprised how many tools produce "you but not really you" results.

The Numbers So Far

2500 users
4.8/5 average user rating
$0 ad spend

All organic growth. Turns out when people get a headshot they love, they tell their friends.

Try It Yourself

If you need a professional headshot but don't want the hassle of a photoshoot, give MyPicNow a try.

Or if you just want to see how your current LinkedIn photo stacks up, I built a free LinkedIn Profile Picture Analyzer that gives you instant feedback.

Fair warning: once you start generating different styles, it's weirdly addictive.

Building AI tools from Barcelona. Also working on Convo, a real-time AI meeting assistant. Say hi on Twitter.

What I Learned Building Two AI Products as a Solo Founder

Markus Kellermann — Wed, 21 Jan 2026 17:25:35 +0000

What I Learned Building Two AI Products as a Solo Founder

I studied economics, worked in VC and consulting, then realized I'd rather ship things than analyze them.

So I quit and started building.

Two AI products later — MyPicNow (AI headshots, 2,500 users) and Convo (real-time meeting AI, 500 users) — here's what actually worked

1. Ship embarrassingly fast

My first version of MyPicNow was ugly. Like, really ugly. But it worked, and that's all that mattered.

I shipped in 4 weeks. If I'd waited for "perfect," I'd still be tweaking buttons.

The feedback from real users taught me more in a week than months of planning ever could.

2. Talk to users constantly

Every feature in Convo exists because a user asked for it — or complained about something.

I don't build what I think is cool. I build what users tell me they need. Then I watch how they actually use it.

The gap between what people say and what they do is where the real insights live.

3. Organic beats paid every time

I've spent exactly $0 on ads for both products.

What worked instead:

SEO pages targeting high-search and low competition keywords
Build strong backlins
Building in public (like this post)

Paid acquisition is a treadmill. Organic compounds.

What's next

Still building. Still learning. Currently scaling Convo and writing about the journey.

If you're building something too, I'd love to hear what's working for you. Drop a comment or find me on Twitter/X or LinkedIn.

Based in Barcelona. Always happy to chat about product, growth, or building from zero.