AI Boss Tools | Diagnostic 11 • Back to the Hub
How much capital does your organisation have exposed to ungoverned AI use? DAIVAR estimates it as a median figure in US dollars, calibrated by sector and country, from twelve inputs about your size, your AI footprint, and your controls. The model is proprietary and runs on our servers; you see the result, not the method.
Banks have measured operational risk for decades, tracking how often losses happen, how large they are, and how controls change both. AI use in most organisations has no equivalent measurement. Tools get adopted and systems go into production without anyone estimating what a bad year of AI-related losses would cost.
DAIVAR applies that operational-risk thinking to AI. It combines quantitative inputs (revenue, headcount, AI usage, systems in production) with a qualitative read of your controls, then calibrates by sector and country to produce your Exposed Capital at Risk: a median estimate in US dollars, the estimated annual likelihood of a material AI loss event, and what drives your number.
Use the estimate the way risk committees use any VaR-style figure: to size the problem, justify the control budget, and compare yourself against a typical organisation of your profile.
Calculate My DAIVAR →The banking standard for measuring operational risk. DAIVAR works in this tradition: frequency, severity, and control-sensitive scorecards.
Published breach-cost research used in calibration, including sector cost patterns and the added cost of ungoverned AI.
The specific weightings, calibration, and blending are proprietary to Adrian Dunkley and run server-side only. No version of the method ships to your browser.
Your median Exposed Capital at Risk estimate, the likelihood of a material AI loss event, what drives your number, a median benchmark for your profile, and a downloadable one-pager report. Free once you enter your name and email.
The twelve inputs take about two minutes, and rough figures are fine because the model is calibrated to work with estimates. Your business figures are not stored; only your name, email, and result tier are recorded when you unlock the report.
Please enter your revenue, employees, AI usage percentage, and systems count.
The DAIVAR model draws on established operational risk methods: frequency and severity modelling, scenario analysis, and control scorecards, calibrated with published breach-cost research. The full method is proprietary, runs server-side, and returns an indicative planning estimate, not audit, actuarial, or regulatory capital advice.
Please enter a valid name and email address.
No spam, ever. Unsubscribe anytime. By continuing you agree to the Privacy & Data Policy.
Calculating…