Eight years building golden cages.

Before Sovrium, I was listed as one of the first French experts in Zapier, automation software for wiring other software together. It belongs to the “no-code” family, a category of tools that lets someone with no programming experience build their own tools by configuring visual interfaces.

That listing let me build a first company, called Essentiel, offering process automation and the development of bespoke internal tools, specific to my clients’ business needs.

Coming from a developer background, I could get my hands dirty and push no-code tools to their limit, filling in what they were incapable of doing with real code and dedicated applications. Using the two together goes by the name low-code. I became a specialist in other tools too: Airtable, Retool, Webflow, Notion, Apify… on top of building backends and APIs in JavaScript.

I always delivered what my clients asked for. That is what makes the rest of this awkward to write.

And still, I ran into plenty of frustrations.

Even using these no-code tools with the rigour of a resourceful, pragmatic engineer, you reach their limits fairly quickly:

I had to urgently rewrite an equivalent system on dedicated applications for several of my clients, a large part of whose day-to-day operations rested on the APIs I had built with Autocode.

I will not hide it: I could happily have done without that. And the realization of how dependent I was on the tools I used, of how fragile the golden cages are, began to set in.

The wheel, reinvented client after client.

Then a question started coming back, project after project.

Working across many tools and many needs, I soon realized that I was reinventing the wheel a good part of the time. The configuration of a low-code information system for one client had to be rebuilt by hand for another, even when 50% of the features were the same.

So how could I easily reuse for other clients what I had already configured? On what configuration model, and above all on what software?

The lead I followed: separate the business from the infrastructure. It is the principle Sovrium rests on today.

So I started trying to conceptualize information-system configurations for Zapier and Airtable: too complex, not ideal, and I was still inside golden cages.

I then tried switching to open-source alternatives, such as n8n and Baserow: more freedom and more control, but a great deal more technical complexity to set up and to maintain.

Still not convinced, I let myself think big: what would be the ideal technology for designing new information systems that answered every problem I had run into?