Qovery

One of the most interesting conversations I had this week was about governance. A platform engineer told me something along the lines of: “Honestly, we already lost control - the worst part is that leadership doesn’t fully see it yet because the problem is mostly invisible.” I think this is where the real tension is starting to emerge - because the shift already happened. PMs are building - Sales teams automate workflows - Marketing ships demos - Non-engineers deploy internal tools - Agents execute tasks continuously. The problem is no longer: “should people build?” but "how do you operationalize all of this safely at scale?" The scary part is that most of this fragmentation stays invisible for a long time: - shadow deployments, - untracked data flows, - personal AI tooling connected to company systems, - apps running outside governance boundaries, - agents interacting with sensitive internal data. In regulated industries especially - I think a lot of companies are underestimating how quickly this can become a real operational, compliance and reputational problem. My feeling more and more lately is that AI-native organizations will not be the ones with the most AI tools used or with the highest adoption but be the ones capable of rebuilding the right governance, and operational layers underneath without killing velocity. /// Exactly why we’re going much deeper on these topics at Qovery lately - we’re seeing more and more companies getting hit by this wave from the bottom up, often long before leadership or security teams realize visibility and governance are already slipping away underneath!

The adoption of Lovable, Claude Code, Bolt, Replit & all these tools is honestly a bit crazy to watch right now. And honestly… the value is real. That’s why this is spreading so fast internally inside companies. You suddenly have PMs building stuff on their own, sales automating workflows, marketing shipping demos, finance teams creating internal tools in a few hours instead of waiting weeks. It’s pretty insane compared to even 6 months ago. But at the same time, we're seeing more and more tension from CTOs, platform teams and especially security teams lately because underneath all this speed, things start becoming very blurry very quickly. Apps connected to company data from personal accounts, random AI connectors plugged into internal systems, zero visibility on where some data goes, internal tools deployed completely outside the infrastructure perimeter… and now people start talking about “shadow AI” the same way we used to talk about shadow IT. The tricky part is that nobody really wants to stop it because the competitive advantage is very real. The business teams pushing this stuff are often right. Especially for regulated companies and enterprise org - you can feel that people are starting to realize there’s a pretty serious governance / compliance / operational problem quietly building underneath. Feels like we’re entering this weird phase where everyone in the company suddenly became capable of building… but most organizations still don’t really have the operational model to absorb that safely. /// Curious how others are handling this internally because we’re seeing this topic explode lately at Qovery . I'm going to organize a roundtable around it soon because honestly the stories and patterns emerging are fascinating. Already have top 5 speakers about this topic - please DM me if you want to be part of it.

Feels like this is exactly where software is heading right now AI massively accelerated the factory. Cursor, Claude, Lovable, agents… software is now being generated faster than most organizations can actually absorb operationally. We’re only starting to realize that the real bottleneck is no longer software creation itself but everything that happens after. CI/CD QA Validation. Release confidence. Deployment orchestration. Runtime governance. All these systems were designed for a world where humans were generating software relatively slowly. Now you suddenly have: → dozens of parallel workflows → non-engineers building internally → AI-generated changes continuously → agents shipping 24/7 We massively upgraded software creation without redesigning the operational system required to safely absorb that level of velocity. Honestly - we’ve probably never seen this level of pull at Qovery around these topics before and it’s hard not to be excited by what’s currently happening!!