What I learned when our joint venture fell apart

Back in September 2023 I announced a joint venture between Stagent and Artwin. Two booking platforms, two founders with over three decades of solo founding experience between them, one ambition to build the best booking and management software in music. I wrote the announcement with real excitement. I meant all of it.

It didn't work out. We spent nearly two years trying to make it work, had a lot of hard conversations, and eventually decided to disentangle. That process is still underway as I write this, which is something I want to be honest about up front. Stagent is finding its next chapter inside Ryde Ventures, which is how I'm writing this post from the other side of a year I don't fully want to repeat but also wouldn't trade.

I want to write down what I learned, because the honest version of the story is more useful than the clean one.

The ambition was the problem, not the enemy

The plan sounded right on paper. Stagent was newer, built on a modern stack, with a growing feature set. Artwin had two decades of codebase behind it and a loyal customer base in a slightly different part of the music industry. Two different platforms, each with their own maturity, their own users, and their own sense of what the product should be. The idea was that we'd merge the customers onto one stack, rebrand everything to Stagent, and ship a better product than either of us could ship alone. Every pitch deck in the history of joint ventures has looked exactly like this.

What the slides never show is how much the details matter. Migrating a customer base the size of Artwin's onto a different codebase is not a feature. It's a months-long rewrite of assumptions. Every flow in Stagent had been designed for how our users worked. Every flow in Artwin had been designed for how their users worked. When you try to reconcile the two, you discover that most of the gap isn't features. It's workflow, vocabulary, and the small defaults nobody ever documented because they were obvious to the people who built them.

The shape of this problem is specific to vertical software. A generic SaaS merging two generic customer bases can usually paper over differences with permissions and feature flags. A vertical product can't, because the differences aren't options. They're the entire reason users chose the product. Two platforms both serving 'the music industry' turn out to be serving two different industries that rhyme. Booking agencies for DJs and artists don't work the same way as agencies for bands. Agents who book electronic acts don't think about tours the same way agents who book instrumental groups do. The invoicing rhythms, the commission structures, the contract conventions, the itinerary layouts, the language on the emails, all subtly different in ways that accumulate fast.

We underestimated every single one of those differences. Individually, each was a day or a week of work. Together, they were a product we would have had to build twice.

Two captains, one ship

The bigger lesson wasn't technical. It was about us.

Marlon and I had both been solopreneurs for most of our careers. I'd been running my own things for almost a decade by the time we started the joint venture. Marlon had been at it for over two. That's a lot of years making every decision alone, owning every call, absorbing every consequence. It makes you exceptional at running your own ship. It does not prepare you to share one.

Nobody tells you this, but the skills that get you successful as a solopreneur actively work against you inside a partnership. You learn to trust your own judgment because nobody else is going to make the call. You learn to move fast because every delay is your fault. You learn to protect the product because it's the only thing that matters. Put two people who've spent their careers building those reflexes on the same ship and what you get isn't double the founder energy. You get two people fighting the same way they've always fought battles, except now the opposing army is each other.

The disagreements that mattered most weren't about strategy or product direction. They were about the shape of work itself. I'd moved to Hong Kong in the middle of the joint venture and was running my half of the operation remotely, across time zones, on asynchronous communication and a lot of trust. Marlon needed a physical base. He wanted people in the same building, working in the same room, running the same daily rhythm. Neither of us was wrong about what we needed to do our best work. We were both wrong about the assumption that it didn't matter. It turned out to matter enormously. A partnership where one founder is on the other side of the world and the other wants everyone within walking distance every day is a partnership that runs out of patience before it runs out of problems to solve. We never fully accepted each other's answer to that question, and in hindsight the rest of the friction was downstream of it.

Shared passion sounds like the perfect foundation for a partnership. It isn't. Marlon and I shared everything we thought would matter. We both cared about music. We both understood the booking industry. We both loved building. None of that helped when the question was how decisions got made on a Tuesday about a feature neither of us fully agreed on. Shared interests don't translate to shared working styles. Nothing translates to shared working styles except the boring, unglamorous work of learning to work together, which is a skill we had both opted out of by going solo in the first place.

I don't say any of this as a criticism of Marlon. Twenty years of running and shipping the same product in this industry is a kind of staying power most founders never earn, and Artwin is the evidence. I say it because it's the part of the story I wish someone had told me before we started. Two solopreneurs is not a partnership. It's two companies sharing a building.

The moment it turned

We tried to make the partnership work through the end of 2024 and into the first quarter of 2025. What finally moved things was me stepping away. In April 2025 I took what we called a time out. I stopped working on the joint venture, gave up my management fee for the duration, stayed available for advice on cycles and bugs but nothing operational, and let the company run without me to see what it looked like when I wasn't in the daily flow. The time out was supposed to be a pause to think. In practice it was the beginning of the disentanglement, because once I was no longer actively running my half of the operation, the question stopped being 'how do we make the partnership work' and started being 'how do we separate the companies cleanly'.

Disentanglement is an uglier word than exit, and it's also harder than it sounds. Separating two companies that have shared infrastructure, shared customers, shared cap tables, shared bookkeeping, and almost two years of intertwined decisions is not a weekend of paperwork. It's months of negotiation with lawyers, accountants, investors, and counterparties on both sides, most of which has been in progress for the second half of 2025 and some of which is still in progress as I write this. Some of those conversations have been harder than I would have chosen. The parts I'm not going to get into here are the parts that are still being worked out, and the parts that are the kind of thing you write about after they're finished rather than during. What I can say is that the easy part of a joint venture split is deciding to separate. The hard part is doing it, and the gap between those two things is wider than anyone starting a joint venture ever expects.

The silver lining, and the reason Stagent still exists at all, is that somewhere in the middle of all of this the right home for it became visible. Ryde Ventures is the AI venture studio I'd known about for years through Lennert, my partner there, who I first met when I was still a teenager pitching a tech idea at his hosting company. Ryde offered Stagent the kind of structure I hadn't been able to build for it alone and hadn't been able to build for it inside the joint venture. A portfolio seat, a team that knows how to operate around a product, patient capital, and the freedom to rebuild the roadmap around what the product actually needs instead of what the partnership could agree on.

Stagent is a portfolio product at Ryde now. The product continues. The team is bigger. The roadmap is sharper. I'm running it differently from how I ran it alone, and differently from how we ran it together, and for the first time in a while I feel like the structure around it matches what the product needs.

What I'd tell someone considering a JV

Four things, in order of how painful they were to learn:

Shared passion is not a partnership. Working styles matter more than shared interests. Before you sign anything, work together on something small and see whether you enjoy making decisions with the other person. Not whether you agree. Whether you can disagree and still ship. A weekend hackathon is a better due diligence exercise than a dozen pitch meetings.

Vertical software does not merge. If both products serve the same industry, the odds that they actually serve the same users are lower than you think. Map the workflow differences before the migration, not during. If you can't write a one-page document describing how the merged product will work for both customer bases without a lot of 'well, we could add a toggle' sentences, the merger isn't ready.

The shape of work matters more than you think. Remote versus office, async versus synchronous, one base versus two time zones. These look like logistics. They're not. They're load-bearing assumptions about how decisions get made, how trust gets built, and how two founders stay aligned when nobody's in the room. Figure out whether your working shapes are compatible before you sign anything, not six months in when the thing you can't agree on turns out to be whether you're even in the same room.

The hardest conversation is usually the right one. Both of us knew it wasn't working months before we admitted it. We kept hoping the next sprint or the next conversation would fix it. It didn't. The thing that actually moved things forward was the conversation I was most afraid to have, which is almost always the conversation that moves things forward. I'd have been better off starting it six months earlier. The corollary is that once that conversation happens, the practical work of unwinding the partnership is still most of what's ahead of you, and it will take longer and cost more than the version of it in your head.

The lesson I'm taking into the next thing

I learned that I'm a builder who needs a partner of a specific shape, and that shape is not another solopreneur. It's a structure. A studio. A team with complementary roles and a clear operating rhythm, where decisions have a process rather than a negotiation, and where the people around the product can outlast any single founder's mood. That's what Ryde turned out to be, and it's working in a way my previous attempt at partnership didn't. I needed to try the other thing first to know the difference. Most of the useful things I know, I learned by trying the version of them that didn't work.

I also learned, or am still learning, that the ending of a joint venture is a longer and messier piece of work than the start of one. Nobody writes case studies about the unwinding phase, probably because it's rarely finished by the time the people involved want to tell the story. I'm writing about mine in the middle because the shape of the lessons is already clear even if the final paperwork isn't, and because the honest version of this story is the one I wish someone had told me before I signed the term sheet two years ago.

Some partnerships are meant to be friendships. Some friendships are meant to be partnerships. It takes real work to figure out which is which, and the year with Marlon was the work. I'm not through the whole of that work yet. The part of it I can see from here is the part I'm writing about.