Building a $20k/mo portfolio and finally going all-in on a 17-year-old product
Marcel Fahle started Bold 17 years ago, but it was just a side project until recently. He pivoted, rebuilt it into an AI-native product, and went all in. It's currently bringing in $3k/mo, while the rest of his portfolio brings in roughly $17k/mo.
Here's Marcel on how he's doing it. 👇
Making it work with a portfolio
I'm a software engineer from Germany, based in Spain. 25+ years of building software, most recently as principal engineer at Commercetools until August 2025.
These days, I'm full-time on Bold, a video intelligence platform for coaching programs and training academies. I run it with my cofounder, Rob Hope. He handles marketing and design; I handle product and engineering.
The bet: Coaching businesses spend a fortune producing video content that nobody can find or use, and Bold turns those libraries into AI coaches that answer student questions with timestamped citations from the content.
I'm also doing a few other things:
I'm a partner at FounderWell, a coaching program for SaaS founders.
I run a consulting practice called Sprint Zero
And on the side, I still run Pidro, a multiplayer card game on iOS and Android with 70k players. I'm rebuilding it from scratch with AI doing all the coding, which has been the most fun I've had in years.
I generate around $15-20k MRR in combined revenue. The bulk comes from SprintZero, which varies between $10-15k/month and funds the product work. Bold is early. It generates ~$3k MRR from 3 customers, plus a revenue share from FounderWell — I'm still figuring out its exact amount, but it will likely push us past $5k. Pidro (the card game) sits at ~$2.5k MRR.
Taking the bet
Bold started as a headless video platform I ran on the side. My first customer is still on it, 17 years later. For most of that time, it was just hosting and delivery — useful but unexciting.
What changed was AI. Once we could extract structured intelligence from video and have students ask real questions of the content, the product stopped being plumbing and started being something a coaching business would pay real money for.
I had this realization at SaaS Academy. I wasn't going to sign up — I didn't want to spend the money, but my wife pushed me to go. It changed everything. I was sitting in a room full of exactly the kind of operators we'd end up selling to, watching them struggle with their own video libraries. That's when I started taking Bold seriously.
I left Commercetools in August and went all in on Bold in September. I'd had the entrepreneurial itch for years, but always stayed comfortable in engineering roles. AI made the bet finally worth taking.
Getting started
That's when Bold became real. Two small kids at home, family math to do, but a side project with paying customers and a consulting practice that could keep the lights on. "Going all in" was less a brave decision than the right one given what was already on the table.
I built the first AI version of Bold inside Commercetools. They had a huge library of partner interviews, and I got a chat and RAG layer working over it. Watching someone ask a question and getting a precise, cited answer out of hours of recorded conversations felt genuinely magical. That was the moment I knew this wasn't just a side project anymore.
Real customers signing on pulled the product forward. Viva Tuition came first, an ACCA exam prep program migrating around 2,500 videos onto Bold. HRT University, a clinical training program with 300 hours of content, signed next. FounderWell, a coaching program for SaaS founders, is the one running live today, and where we are finding and filing off most of the rough edges. Each has a different shape of library and a different student base, which has been the best possible forcing function for building something that generalizes.
The stack
Here's the stack:
Backend is Elixir/Phoenix on Postgres, multi-tenant via Triplex with Oban for background jobs and a hybrid keyword + vector search layer. Boring, durable, easy to maintain solo.
Video runs on Mux and S3, with transcription routed across a handful of speech-to-text engines (AssemblyAI, Deepgram, Speechmatics, Sonix, Rev) depending on language, accuracy, and cost. The AI layer is Anthropic and OpenAI, but I'm constantly evaluating other providers and models.
Frontend is a headless Next.js starter. Bold is API-first, so customers can use our starter, build their own, or wire Bold into an existing site. The fast-launch combo we recommend is Bold + Sanity for CMS + Outseta for membership.
Hosted on Gigalixir today, evaluating a move to Hetzner + Kamal. Indie founder math.
The hardest part wasn't the code. It was the gap between "this works for one customer" and "this is a product." That meant rewriting the same systems three or four times until the patterns stopped breaking when a new customer showed up with their own oddly-shaped library.
Admit that you're hedging
The biggest obstacle was me. Not the market, the tech, or the customers.
For years, I had a comfortable engineering job, a working side project with real customers, and every excuse I needed to keep one foot in each world. The hard part wasn't building, it was admitting I'd been hedging — that I'd built up a very polished story for why "now isn't the right time." I did that for longer than I'd like to admit.
It was impostor syndrome. I've spent most of my time around brilliant people, and the reflex is to assume everyone else has figured out something you haven't. So you stay quiet.
Customer feedback broke that — and Brene Brown :)
When someone running a 2500-video training program looks at what you built and says "this is what we've been waiting for," it gets harder to keep believing you don't belong in the room. I still feel it though!
If I had to start over, I'd grind my face off much earlier. I'd do the inner work, name the procrastination for what it was, and just go. Everything I'm doing now I could have been doing three or four years ago.
Inbound leads
Bold is a straightforward SaaS. We price it based on library size, active students, and AI usage. Customers pay between $700 and $1,500 per month; an enterprise tier is available above that. We offer no free or low-tier plan and do not plan to, because the intelligence layer doesn't earn its keep below a certain scale.
Our growth has been entirely inbound: warm intros, peer networks, founder-to-founder. Much of it has come through founder coaching circles, which also led to my partnership with FounderWell. Our Ideal Customer Profile (ICP) almost perfectly overlaps with those already in these circles.
No formal playbook exists yet. The first three customers came because the product did the talking, and word travels fast in coaching circles. At this stage, I'd rather have one customer obsessed with the product than a hundred who showed up because of a clever campaign.
That said, we are starting more deliberate work now. Rob's leading a content push; I'm doing more outreach and showing up at events, etc.
Get started by breaking tasks down into micro tasks
I have two pieces of advice, both borrowed from trail running, which our family obsesses over.
Break complex things down until each step is small enough to start. Most "I don't know where to begin" moments aren't about not knowing. They're about the next step being too big. Cut it in half, then cut it in half again, until what's in front of you is something you can do in the next hour. Then do that.
And when you fall off, don't make it mean anything. No judgment, no story about why this proves you're not cut out for it. Just begin again. The founders and runners who go the distance aren't the ones who never miss a day. They're the ones who don't turn a missed day into a missed week. That single skill, the ability to start over without making it a referendum on yourself, is most of the game.
What's next?
I want Bold to be a great business that solves a real problem. Coaching programs contain enormous amounts of expertise locked in video, and most of it remains unused. Closing that gap matters.
I am building toward a small, high-quality team, customers who genuinely love the product, and a company that supports my family without being a treadmill.
Not the biggest exit or the most logos. Something durable, where I'd be happy to still be doing the work in 20 years.
You can learn more at:
Leave a Comment
Really appreciated the insight on how your growth has been entirely inbound via warm intros and coaching circles. As a developer building a portfolio of niche tools, I’ve found that being active in founder communities is far more effective than cold outreach. It’s about building trust first. Thanks for sharing the breakdown of your revenue mix
One of the most interesting parts here is the patience behind compounding small wins over a long period of time. A lot of founders underestimate how powerful consistency becomes when paired with deep product familiarity.
Going all-in after 17 years also says a lot about conviction and timing rather than chasing short-term momentum.