News
Kyvvu raises a €1M pre-seed from Volta Ventures and BOM
We raised a €1 million pre-seed round from Volta Ventures and the Brabant Development Agency. What the round pays for: the team, the partner network, and security accreditation.
Maurits Kaptein
· 2 min read
We raised a €1 million pre-seed round from Volta Ventures and the Brabant Development Agency (BOM). The announcement went out on 21 August and was picked up by IO+, Dutch IT Channel, Dealroom and JADS.
What we are building
An agent carries out a task as a sequence of actions. Each action can be permitted and the sequence can still be a breach: read a customer record, then send an email outside the organisation. Both steps pass their own check. Together they are a leak.
Identity systems check one action at a time. Content guardrails check one message at a time. A sandbox bounds where an agent can act, not the order in which it acts there. None of them can ask: given everything this agent has already done in this task, is this next action allowed?
That is the question Kyvvu’s Agent Security Kernel answers — a check that runs inside the agent’s own process, before the action runs. The reasoning is in our white paper and in Runtime Governance for AI Agents: Policies on Paths. The kernel is in use at financial services, insurance and healthcare organisations in the Netherlands, together with our integration partners.
Where the money goes
The team. We expanded engineering after the round closed and will keep adding people. The other two depend on it.
The partner network. Kyvvu reaches production systems through the parties that build and run agents for regulated organisations. Widening that network puts the kernel in more places than we could reach ourselves.
Security accreditation. Banks, insurers, healthcare providers and public bodies ask for evidence before a pilot becomes a deployment. ISO 27001 and SOC 2 work is underway.
What does not change
Kyvvu runs inside the agent. There is no proxy, no gateway, nothing your traffic passes through, and no sensitive data sent to us for checking. The SDK stays Apache-2.0 and the engine stays source-available, so you can read the code that enforces your policies. The Agent Action Grammar stays open and vendor-neutral, with RFC-0001 open for comment.
If you want to work on this, our open roles are here.
Keep reading
- blog#dora
· Maurits Kaptein
What does DORA ask of your AI agents?
DORA does not explicitly mention AI, yet every agent a bank or insurer builds, buys or hands to its developers is an ICT system to which DORA applies. In this post we try to analyze the DORA requirements and highlight how they relate to agent governance and security.
- blog#agent-security
· Maurits Kaptein
We Used to Be the Only Ones. We're Not Anymore.
A reflection on OWASP’s Agent Control Standard, Microsoft’s Agent Control Specification, and AGNTCY’s SHADI, seen through the lens of a reference monitor — and how SHADI extends what Kyvvu covers.
- news#kyvvu
· Maurits Kaptein
kyvvu-claude: policy enforcement for Claude Code, and a beta for coding agents
kyvvu-claude checks every Claude Code tool call against your policies before it runs, using Claude Code’s own hooks. Claude Code works today; Cursor, Copilot, Codex and OpenCode are being built with the beta teams.
Kyvvu
Every agent has to ASK before it acts.
See what Kyvvu blocks on your own agents, or put one under enforcement yourself in five minutes.