My 2025 Retrospective

Happy Saturday!
(Yes. After 2+ years of Friday emails, I'm making the switch to the weekend. Also writing this between Lego builds with the kids, so if I forgot something, blame the Plane and Airport Terminal.)

Overall, it's been a helluva year. I hit 300% of my revenue goal. Usually, that's the signal to double down.

But I'm doing the opposite.

Because looking back at the work, I realized I was too often giving maps to people who needed to start driving.

A year ago, I published a plan to help 100 people add $10K profit to their business with AI. Six parts. Daily emails. Biweekly workshops. A new membership. The whole thing built around one simple idea: stop chasing "cool AI" and start chasing profit.

Now it's time to look back at what actually happened – and what's changing.

Let's dive in.

The Plan I Made

In January 2025, I wrote a post called "My Plan To Help 100 People Add $10K Profit To Their Business With AI."

The rules were simple:

• We don't care if it's the latest and greatest – we care if it makes money

• We don’t want to make a big upfront investment

• Every new AI project has to pay for itself in the first 3 months

• We unlock bigger projects by taking lots of smaller wins in a single direction

The plan had six parts: more content, daily emails, biweekly workshops, 1:1 work through the AI 10K, classical consulting for select clients, and an "unlimited advisory" offer I'd never tried before.

The goal was to help 100 people hit that $10K mark.

Let’s see what happened actually.

The Numbers

127 new customers

And that number doesn't even count workshop tickets. So technically, I exceeded the "100 people" goal.

But let me be honest: I can't prove that all 127 hit $10K in profit. Some did – clearly. Others got value in ways that are harder to measure. What I do know is that people stuck around. The AI 10K Club now has over 70 active members. Many renewed on Black Friday for another year.

That tells me something worked, even if the scoreboard is fuzzy.

Revenue split into three roughly equal parts:

1) Corporate Consulting (~1/3)

I worked with 5 major clients this year (4 from Germany, 1 from North America) plus a handful of smaller gigs. Revenue was up about 30% compared to last year. I cap most engagements at 5-10 days per month, so growth here is limited by design.

2) AI 10K Club + Workshops (~1/3)

This was completely new. Zero revenue from this stream in 2024. Now it's a real business line. Workshops, membership, tools — all built from scratch over the past 12 months. I’m loving this one and the Club is showing nice steady growth over time. One-time workshop passes sold another 200+.

3) Courses + Speaking (~1/3)

I did WAY more courses than planned. One, because there was a market need, but the other big factor was that producing courses with platforms like LinkedIn Learning and O’Reilly was really fun.

It's always a great experience to at the LinkedIn Studios in Carpinteria, and this year I went there twice. 25,000 people already took my courses on LinkedIn Learning.

And my O’Reilly LIVE trainings consistently brought me in touch with learners from all around the world in a super intense setting (the longest session was 5 hours live and we had 100+ people stick around!).

The course business brought in easily 4x what I expected. Speaking gigs were fewer, but intentionally so. I limited myself to 1-2 per quarter.

While doing all that, I also published my third book — "The Profitable AI Advantage" — that hit #11 on Amazon's AI Books Bestseller list. And somehow ended up meeting Jensen Huang at a private Nvidia event in Paris. Hard to put a number on that one.

The 300% growth in a nutshell: AI 10K was entirely new. Consulting grew steadily. Courses overperformed. But the real engine was building something from nothing.

The Wins

Let me share a few concrete examples (anonymized, NDAs and all that).

One enterprise client committed to 15+ chatbots this year after deploying the first 5 and realizing that they saved thousands of hours in first-level support (which turned out to be a major bottleneck since they’re in the events industry). Real workflows, real savings, real adoption.

Another client needed help seeing the full picture (mid-sized organization). We started systematically mapping their entire AI opportunity landscape and prioritized using the $10K threshold. The roadmap is now looking at $1M+ in efficiency gains and is being implemented as we move into 2026.

On the other end of the spectrum: a solo founder who was paying contractors for work AI could handle. We built the workflows together. Now they're insourcing and keeping the margin which means $20K+ per year, back in their pocket.

And then there was a young healthcare startup. One AI workflow replaced what would've been a full-time hire just for data collection.

Big and small. Both work.

What I Learned

1) AI consulting is a loop

You end up having the same "no, this won't work like that" conversations. It’s repetitive, but necessary. AI has nuance and movement creates friction.

Adding AI doesn't simplify things at the beginning. It adds complexity. The learning path isn't linear.

If you learn carpentry, you can build a solid shelf. If you learn "AI basics", you still can't build a great AI solution.

There are too many moving parts: data, workflows, models, tooling, edge cases, organizational constraints. Knowing the basics doesn't translate directly into outcomes.

In that mess, guidance is the only shortcut.

2) People want proven tools and tutorials – not more options

I bet on customization. But the market wanted structure.

Step-by-step. "Do this, then this, then this."

Even when their problems are unique, they'd rather start with a template and adapt it than face a blank canvas.

Structure beats freedom, especially at the beginning.

3) Roadmaps create paralysis. Production creates momentum

This is the big one.

If I look at two businesses:

• Business A has a crystal-clear AI roadmap but nothing in production.

• Business B has a messy strategy but one profitable AI workflow.

Business B wins. Every time.

Because more possibilities create more paralysis. And successful production creates momentum.

Once you have one thing running, building the next becomes easier. You've proven it works. You've built the muscle. You've earned the credibility internally.

But if you're stuck in the planning phase, you never get there.

4) "AI First" can be a trap

Companies proclaim they're going 'AI First'. The big transformation. Customer service, ops, marketing, data – everything, all at once.

So they plan. They build a comprehensive roadmap. They hire consultants to map the territory.

And they ship... nothing.

"AI First" sounds bold, but it often leads to transformation theater.

What actually worked consistently is a "First AI" model.

Build one relevant, engineered AI solution that works profitably. Find one messy, ugly workflow that saves or makes $10K. Build it. Ship it. Break it. Fix it.

Once that first one is running, the roadmap becomes much easier.

You don't need to see the whole staircase to take the first step. You just need to see the step.

5) Measurability is hard.

This is something I'm still wrestling with.

I learned the hard way that "Productivity AI" (ChatGPT, Copilot, general assistants) and "Engineered AI" (custom workflows, integrated solutions) need different rules.

People "feel" that Productivity AI is valuable. But proving $10K in savings? Nearly impossible for most use cases.

Engineered AI is different. You can measure throughput, time saved, error rates. The ROI is tangible. But so are the costs.

I'm going to be much clearer about this distinction going forward. Each has its place, but you can't measure them with the same ruler.

What I'm Doing Differently in 2026

AI 10K gets narrower.

The original AI 10K could feel overwhelming. Too much possibility. Too many decisions.

Going forward, the goal is simpler:

→ Find one profitable AI opportunity worth $10K+ and then build it.

Not ten ideas. Not a master plan. One concrete opportunity with real, relevant upside.

The roadmapping part still makes sense, but mainly in corporate 1:1 consulting, or direct work with me when you want my hands on the keyboard.

For everyone else, I'm building tools and templates to help you nail that first win without needing a comprehensive strategy.

More building, less theory.

The best workshop feedback consistently came from the "easy" stuff. Building AI workflows in n8n, hands-on prompts, practical templates, etc.

Less theory, more shipping.

I'll do more of that.

The bigger vision.

Long-term, the goal is to make myself obsolete.

Not disappear — but package what I do into smaller, usable pieces. "Mini versions of me" that help with the internal work: finding the right problem, setting your threshold for "what's worth building," mapping AI capabilities to real opportunities.

I have an idea brewing. Something that could automate a lot of what I currently do manually.

I'll share more when it's ready.

My big newsletter moves to Saturday.

Small, but intentional change – the last Fridays often felt rushed. Saturday gives me (and you) more breathing room. Apart from that, I’m still sending my daily profitable AI notes (1-minute reads):

The Bottom Line

In 2025, I focused a lot on the staircase. Helping people see the full picture. Building roadmaps. Mapping possibilities.

And I still do that – when it's what your business needs.

But in 2026, I'm paying even more attention to the first step.

Because the real metric isn't "AI adoption". It's the number of profitable AI solutions you're running, and how fast you got the first one.

I'll be in your corner to accelerate that journey.

See you next Saturday,
Tobias