API Evangelist Conversation with Zdenek Nemec on What Superface Taught Him, Autonomous API Onboarding, and Why Transactional APIs Go First

I learned the most about myself — my personal traits, where I might be irritating people or not getting along with them. That is the biggest and deepest learning, and it is not in the write-up. But a couple of things I already knew got reinforced. Deep tech cannot really be done from Europe. Apiary kind of did it, but we were in San Francisco. Maybe around COVID there was a short window when it was possible, but for deep tech you need to be in the Bay Area. And timing. General Magic is my favorite documentary — timing, distribution and network matter more than any idea or technology. That is now the prism I look through for anything I am thinking about doing. Do I have the distribution? Do I have the network? Is this the right time, the right market, the pain people have today? I knew these things before, but I chose to ignore them.

Wrong time and wrong place. I had a vision of a future I wanted to bring about, but it was poorly executed, at the wrong time, in the wrong place. Coming out of COVID, the money shifting, the AI shift, social media shifting — you are right that it was a much harder moment to ignore timing, distribution and network than it would have been a few years earlier.

Technically it is possible to some extent, but it is still one of the biggest hurdles in autonomous integration. I always saw it as the critical blocker for my vision of autonomous APIs — software that can integrate on its own, purchase access to capabilities and execute. The real question is whether there is an economic driver behind it. We now have new actors, agents, that may create that business need, and I expect it will come. I think a lot about the economic drivers for APIs. Direct monetization was only ever a tiny fraction of why companies built APIs — maybe ten percent, the Clearbits and the send-a-text-message transactions. Onboarding into those transactional APIs is technically, legally and commercially straightforward, and with x402 and MPP it should become the norm.

Onboarding into a CRM like Salesforce, or ordering a physical service — asking DHL to ship six containers from Prague to New York. I do not see that in the next several years. Maybe someday a bot will do it, but even as I am saying it, I do not believe it. I do not see an agent onboarding itself into Oracle Hospitality Cloud, even though that is exactly where it would make sense — book this room. That is not happening soon, and by soon I mean years. What I would love to see is a prediction for the transactional side. Before 2020, direct monetization was a very marginal part of the API economy. With these new actors, how does it grow? I do not think anyone is really talking about it, and I would love to see someone do that research.

Right before this call I talked with a younger founder building something close to my old vision — he hated me calling it a marketplace of capabilities, but it is discovery and purchasing of capabilities. When I challenged him, he said the API economy I was describing belongs to the present and the past. People who started programming in a world of AI and agents will not build a new CRM or system of record the way we did. They will not start with developer portals, API gateways and complex relations between resources. They will start with autonomous onboarding, direct monetization and discoverability as the default. And I kind of like it. The next viral project, like OpenClaw, will come with this built in and set the trend, and as more people build new APIs that way, one way or another we will get there.

I find it very ironic. Amazon, the company that brought e-commerce to the world — and destroyed brick and mortar, if you will — is now afraid it is going to be destroyed by agentic commerce. The same pattern is playing out again, just with Amazon on the other side of it this time.