Why we built an independent AI firewall
Lakera, Protect AI, Prompt Security — all acquired in 2025. Here's why we think the market needs an independent, European alternative.
The market consolidated faster than anyone expected
In 2025, three of the most significant LLM security companies were acquired. Lakera by Cisco. Protect AI by Palo Alto Networks. Prompt Security by a US defense contractor. In twelve months, the independent LLM security landscape went from vibrant to nearly nonexistent.
For large US enterprises, this is fine. They already buy from Cisco and Palo Alto. Adding LLM security to an existing contract is a procurement no-brainer.
For European startups, scale-ups, and mid-market companies? It's a problem. Hosting your security infrastructure in the US is increasingly difficult to justify to legal and compliance teams. Especially with the EU AI Act arriving in August 2026.
The EU dimension
The EU AI Act isn't optional. If you're deploying AI systems in the EU that interact with people — and an LLM assistant does — you need an audit trail. You need data classification. You need to demonstrate that you've thought about risk.
The US-based security tools weren't built with GDPR and the EU AI Act as first-class constraints. They were bolt-ons. Compliance reports generated after the fact, not woven into the architecture.
Senthex is built from the ground up for EU compliance. Hosted in Germany (Hetzner). Zero data retention by default. The audit trail is a byproduct of how the system works, not a separate product you pay extra for.
Independence is a feature
When your security vendor gets acquired, things change. Pricing goes up. The roadmap shifts toward the acquirer's priorities. The small-team features get deprioritized. The startup-friendly pricing disappears.
We're a small team building something specific. We're not planning to sell. We're planning to build the best LLM security tool in Europe and run it as a sustainable business.
That independence is part of the product. When we say our pricing is startup-friendly, we mean it — and we can keep meaning it because we don't have a US acquirer's margins to justify.
The one-line bet
The other strategic choice we made early: zero integration friction. One line of code. If we make developers jump through hoops, they don't adopt. If they don't adopt, the risk stays. Simple.
We watched teams evaluate complex security tools and abandon them in the POC phase because the integration was too heavy. We built the thing we wished existed — the thing where you could be protected before your coffee was done.
That's Senthex. Independent, European, and unreasonably easy to deploy.
By Yohann Sidot