Soverli founding team
[ 01 ] Mobile Security Seed

Soverli

Sovereign smartphone architecture

Mobile Cybersecurity Sovereignty

Building a sovereign smartphone architecture that runs fully-isolated operating systems alongside Android and iOS on commercial devices — delivering auditable, sovereign security without sacrificing mainstream usability.

Mobile security has long demanded a binary choice: locked-down dedicated devices or the convenience of mainstream platforms. Soverli eliminates that tradeoff. The founding team brings deep OS-level and cryptographic expertise to a problem that government and enterprise customers cannot avoid.

Stage
Seed
Domain
Mobile Security
Geography
Switzerland
Visit website
CyberDesk founding team
[ 02 ] Authorization Seed

CyberDesk

Authorization layer for human, non-human, and AI identities

Identity & Access Enterprise AI-ready

Building the authorization layer that removes standing access to data by continuously enforcing who can do what, on which data — across human, non-human, and AI identities.

The shift to AI agents in the enterprise creates a new class of identity problems that existing access controls were never designed for. CyberDesk's continuous enforcement model treats every access decision as a fresh evaluation — a technically sound answer to a problem that is becoming acute for every regulated business.

Stage
Seed
Domain
Authorization
Geography
Germany
Visit website
Refractal founding team
[ 03 ] AI Security Pre-Seed

Refractal

Runtime security for AI agents

AI Agents Runtime Security Enterprise

Building the security layer that sits between AI agents and the tools, data, and people they touch — checking every agent action against policy before it runs, and allowing, blocking, or redacting it in real time.

Large general-purpose models are too slow and too costly to sit in the hot path of every agent action. Refractal trains compact security models on each customer's own policies, workflows, and telemetry, deployable in their cloud or fully on-premises. Before going live, the models are tested against simulated replicas of customer environments, and uncertain decisions are routed to a human reviewer — a design built for regulated teams working under the EU AI Act, GDPR, and DORA.

Stage
Pre-Seed
Domain
AI Security
Geography
U.K.
Visit website

We are deliberately selective. A small portfolio means meaningful time with every founder — not a scorecard to optimize.

IYK is focused on being helpful, not on being additional work.

Tobias, Co-founder & CEO, CyberDesk

Building something that belongs here?

We're always open to conversations with technical founders at the edge of cybersecurity and digital infrastructure.