Slush 2026 brought more than 12,000 founders and investors to Helsinki for two days. The organisers called it a bet on Europe. From where we stood, it looked more like a bet on AI agents doing a big share of Europe's work.

That's a bigger shift than it sounds. Agents no longer just answer questions. They send emails, merge code and move money, often with broad access and no clear owner. Which brings us to the question we work on every day at Cygrid: when software starts acting on its own, who decides what it's allowed to do?

It's the question our Founder & CEO, Dmitry Kharchenko, brought on stage with him.

"AI went from talking to acting"

Dmitry had been asked to explain one thing: why is now the right moment for Cygrid? His answer was short. AI went from talking to acting, and nobody controls what it does.

Then he explained how fast it happened. In about 18 months, AI went from drafting emails to sending them, from suggesting code to merging it, from answering questions about invoices to actually moving money. Around the same time, the Model Context Protocol (MCP) made it possible to plug an agent into almost any SaaS app in minutes. And while all of that was happening, NIS2, DORA and the EU AI Act made companies answerable for what their systems do, whether a person pressed the button or not.

Put those three things together and you get a gap. Identity security was built for humans logging in. Nobody built it for software that acts on its own, around the clock, across dozens of tools. "The agentic enterprise needs a new control layer," as Dmitry put it, "and Cygrid now sets the standard for the next decade. The window is open, and we're already through it."

Once you hear it framed that way, you start noticing it everywhere. And at Slush, it was everywhere.

The agents are already at work

Look at who was speaking this year: people from Lovable, n8n, Cognition and Anthropic, all building software that writes and ships code, runs workflows and works across other tools on someone's behalf. Only a couple of years ago, AI on stage meant benchmarks and chatbots. Now it means software with a job.

That's exciting. It's also a little uncomfortable if you're responsible for security. Think about what happens when a new employee joins. They get an account, a manager and access that matches their role. When they leave, someone switches it all off. An agent that approves payments is doing the same kind of work, but it usually arrives with none of that. It has the access, and nobody is quite sure who's responsible for it.

The answer isn't to slow agents down. It's to treat them like the colleagues they are becoming: give each one a named owner, access that fits the task, and a way to stop it when something looks wrong.

Europe wants to know who's in control

The same theme showed up on the policy side, just in different words. One of the speakers was Henna Virkkunen, whose job title at the European Commission is Executive Vice-President for Technological Sovereignty, Security and Democracy. Sovereignty and security, in a single portfolio. She shared the speaker list with Commissioner Ekaterina Zaharieva and Finland's President Alexander Stubb, while EU-INC, the push for a single European company structure, seemed to come up in every other conversation.

For a security team, sovereignty sounds abstract until a customer or auditor asks the practical version. Where is our data processed? Which jurisdiction can demand access to it? Can you show us who touched it, including your AI? Those questions were already hard when only people moved data around. With agents moving it between more systems, faster, they become harder still, so it's worth making sure you can answer them for your AI tools and not just your core systems.

Resilience has become part of the pitch

Defence and deep tech had a real presence too, with founders from Helsing and ICEYE among the speakers. You could sense a change in mood. Resilience, including cyber resilience, is no longer treated as a cost to be kept small. It's increasingly part of what makes a European company worth backing.

Regulation got there first. DORA has applied to EU financial companies since January 2025. NIS2 is bringing security obligations to many more sectors and holding management accountable. The EU AI Act is phasing in rules for record-keeping and human oversight of AI. Different laws, one shared demand: show who had access to what, and prove your controls actually worked. A spreadsheet updated once a year can't do that when agents act thousands of times a day. The records need to be continuous, and ready whenever someone asks.

Small teams, lots of identities

Slush is, above all, a founder event, and a lot of the conversation was about building more with fewer people. That's great news for productivity. It also quietly creates a lot of identities.

Picture a typical 20-person startup. Every engineer connects a coding assistant to GitHub. Sales adds an AI note-taker to calendars and the CRM. Operations sets up a few automation workflows, each with its own API key. A few weeks later, the company has more non-human identities than employees, and hardly any of them have a named owner. None of that is a reason to slow down. It's a reason to get the basics right early, while knowing what's connected and cleaning up unused access is still a job for an afternoon rather than a quarter.

What we brought home

We left Helsinki convinced of one thing. AI agents are already at the centre of how companies work, and the companies that pull ahead will be the ones that move fast with AI and stay in control of what it can reach.

If you want to know where you stand, three questions are a good place to start:

  • Do we know which AI agents are connected to our systems, and who owns each one?

  • Can we stop a risky action while it's happening, not weeks later?

  • Can we show where our data is processed and who accessed it, including AI?

If any of those made you pause, you're in good company. If we met at Slush, thank you for the conversations. If we didn't, let's talk.


Masha Kovaleva

PR & Marketing Manager

Share

Stay in the loop

Product news and identity security insights.

Platform

Solutions by Team

Solutions by Industry

Company

Legal

© 2026 Cygrid GmbH. All rights reserved.

Stay in the loop

Product news and identity security insights.

Platform

Solutions by Team

Solutions by Industry

Company

Legal

© 2026 Cygrid GmbH. All rights reserved.