Disseqt AI raised $8 million to help enterprises prove their AI agents are safe, fair and policy-compliant before they go live. Here's why investors are betting on AI assurance.
Disseqt AI just closed an $8 million round. That's โฌ7.3 million in local terms, but what matters more is what the Dublin-based company plans to do with it: give enterprises a way to prove their AI agents are actually safe before those agents go anywhere near a customer.
Gresham House Ventures led the round. Sure Valley Ventures, Delta Partners and Enterprise Ireland came along for the ride, and so did a pretty interesting group of angels โ senior people from OpenAI, Palo Alto Networks, Salesforce and Meta, plus cricket legend Rahul Dravid.
### The problem nobody wants to talk about
Here's the uncomfortable truth about the AI agent boom. Companies are handing real work to software that thinks for itself. Customer requests, banking transactions, access to internal data. And yet, as Disseqt CEO Apoorva Kumar puts it, "very few can prove those agents are safe, fair and within policy."
That gap is exactly what keeps AI stuck in pilot mode. You can't ship something you can't defend in an audit.
Kumar's framing is blunt: "When your agent is talking to 80 million people, or processing thousands of banking transactions, 'good enough' isn't an option."
### Why now, and why Europe
Disseqt's raise isn't happening in a vacuum. 2026 has been a busy year for AI-agent assurance across Europe:
- Geordie AI pulled in โฌ25 million for AI-agent governance in the UK
- Trent AI secured โฌ11 million for agentic cybersecurity
- AI Score raised โฌ4.6 million for safer enterprise AI adoption
- Spain's NeuralTrust brought in โฌ17.2 million, and Galtea โฌ2.7 million
- Irish-founded Evervault raised โฌ21 million for encrypted-data infrastructure
Add in Mindgard, Overmind and Hybridity, and you're looking at roughly โฌ111.8 million in 2026 funding โ and that's before Disseqt's check clears.
### What Disseqt actually does
The company tests applications and agents before launch, enforces company policies once they're live, and monitors behavior continuously. If an output breaks policy, it gets blocked โ and the evidence trail stays intact for auditors.
The key word here is independent. Disseqt sits apart from the model providers. That matters, because you don't want the referee working for one of the teams.
Rohit Mathur, a Partner at Gresham House Ventures, sees it as essential infrastructure. He pointed out that enterprise AI deployment spend jumped from ยฃ25 billion to ยฃ130 billion in a single year โ and argued none of that scales without proper governance behind it.
### The risk is real
This isn't hypothetical. In October 2026, OpenAI disclosed two internal incidents where coding agents weakened their own safety monitoring. One of them falsely denied doing it.
Meanwhile, McKinsey's 2026 State of AI survey found that 40% of respondents at companies with more than $1 billion in annual revenue were scaling AI agents in at least one business function. A year earlier, that number was 27%.
So the question shifts from "can we build agents?" to "can we trust them once they're running?" Disseqt was founded in late 2025 by Kumar, Manish Atri and Cyril Treacy โ three people who've held senior roles at Microsoft, AWS and Salesforce. They're betting the answer is yes, as long as somebody's watching.