It’s been around 7-8 months since AI Dubliners first featured Safentic. In the AI world, that’s a long time.
Back then, agentic AI still felt largely like a promise.
Today, AI agents are scheduling meetings, sending emails, accessing company systems, deleting files, triggering workflows, and making increasingly autonomous decisions.
📉 Deployment cycles have accelerated. 🎯 Expectations have changed. 🚀 Capabilities have expanded.
But one question remains the same: What happens when an AI agent makes the wrong decision? In fact, that question feels more important today than ever before.
Many organisations are now focused on how quickly they can deploy AI agents. Safentic is focused on a different challenge: How safely they can deploy them.
That’s why it was great to see recent milestones from this Dublin-based startup:
-Founder and CEO Marjia Siddik became Ireland’s only Microsoft Trustworthy AI MVP and one of just 28 MVPs globally.
-Safentic was selected as an Onstage Top 100 startup from more than 1,500 applicants.
What makes Safentic particularly interesting is its approach to AI safety. Traditional monitoring systems often detect problems only after the damage has already been done.
Safentic operates at the decision layer:
🔹 When an AI agent attempts to access a tool.
🔹 When it wants to make an API call.
🔹 During the decision-making process before an action is taken.
It evaluates policies, records decisions, and can intervene when behaviour violates defined safety rules.
As AI systems gain more autonomy, governance is gradually shifting from a compliance discussion to an operational necessity.
Dublin has no shortage of companies helping organisations move faster with AI. Safentic is asking a different question: Before we move faster, can we make sure we move safely?
It was an important question 7-8 months ago. Today, it may be more important than ever.
