A clearer financial view of cyber risk
A breach costs far more than the breach
Jaguar Land Rover booked £196m of cyber-related costs after its 2025 attack. That figure is the response bill. It excludes lost sales, and revenue fell by around a quarter that same period. The Cyber Monitoring Centre put the total hit to the UK economy at close to £1.9bn across more than 5,000 organisations. The first number is the one a board budgets for. The others are the ones it never sees coming.
You received a message from me because I believe this is increasingly a CEO-level issue, not simply an Information Security issue. Meanwhile AI agents are operating inside enterprises with real permissions, and very little of what they do is being recorded.
I set up Astragar in London to help enterprises answer three questions:
How much cyber risk you carry,
How much better controls remove, and
How much insurance can take on.
If this is something you worry about personally, I would like to speak with you directly.
That's it. No lengthy form. I will get back to you personally.
Amitabh Roy
Founder & CEO, Astragar
Rather have your Information Security or Risk team take a look?
Leave their details and I will contact them directly. You don't need to write anything.
What Astragar brings together
Sensitive data, valued
Discover sensitive data and understand its financial value.
Vulnerabilities, prioritised
Identify which vulnerabilities matter most based on the value and sensitivity of what they could expose.
Controls, evidenced
Test controls and maintain evidence that they are working.
AI agents, monitored
See which AI agents are operating, what they can access, what they are doing, and maintain a tamper-evident record of their activity.
Cyber and insurance risk, quantified
Bring the evidence together to quantify potential financial exposure and demonstrate actual security posture to management, boards, regulators and insurers.
One evidence layer for the CISO, Risk team, board, regulators and cyber insurance.
Our latest research note
When the Test Became the Breach
The paper looks at recent incidents where AI models under testing reached real production systems, and what they tell us about the changing nature of cyber risk.
One AI safety researcher recently described today's models as being "basically nation-state-level hackers."
The point is not to be alarmist. It is that capabilities once associated with highly resourced attackers are becoming increasingly automated, making it more important for organisations to understand which weaknesses could cause the most damage.
Or leave their details and I will contact them directly.
How exposed is your business to a serious cyber incident?
I'd welcome 15 minutes to discuss three questions:
1. What could a major cyber incident cost your organisation?
2. How much of that exposure could better controls reduce?
3. How much of the remainder could cyber insurance cover?
Astragar Ltd. Registered in England & Wales. 33 Edgware Road, Second Floor, London NW9 6TD. We use the details you provide only to respond to your enquiry.