• Raised €2.3M to build a supervision layer for AI agents — selling seatbelts while everyone else races to build faster cars.
• Four employees trying to monitor every AI agent in regulated industries. That's ambitious or delusional, depending on the quarter.
• Targeting EU AI Act compliance 18 months before anyone knows what compliance actually means.
• 'AI model sovereignty' — finally, a startup pitching geopolitical risk as a feature, not a bug.

2-4 years before OpenAI or Anthropic ships native observability that's 'good enough' for 80% of use cases. Your window is the regulatory chaos.

• 'Continuous improvement through reinforcement learning' — every ML startup's pitch deck since 2018, now applied to watching other AIs.
• 'Ownership of AI models and infrastructure to ensure sovereignty' — pivoting Brexit energy into B2B SaaS.

If Overmind was a person, it'd be that friend who became a lifeguard right as everyone started jumping into the deep end — needed, but perpetually overwhelmed.

• Regulated industries genuinely need this — legal/healthcare/fintech can't YOLO deploy agents without audit trails.
• First-mover advantage in a category that doesn't exist yet. EU AI Act is vague enough to be your moat for 24 months.
• Osney Capital backing signals sector credibility — they don't fund random AI wrapper companies.
• Competitors (Protecto, Borneo) are equally small and equally early. Land 3 enterprise logos first and the space is yours.

You're building a supervision layer for AI agents that will eventually supervise themselves. It's turtles all the way down, but venture-backed.

• Ship a free tier for startups NOW — become the default observability layer before Anthropic builds this into Claude Enterprise.
• Pick ONE vertical (legal or healthcare) and get 5 public case studies by Q3 2025. Generalist positioning is death at 4 employees.
• Hire a ex-Big4 compliance person who can translate 'EU AI Act Article 14' into sellable features. Regulation is your moat — weaponize it.