Your attack surface grew in two directions at once, and the half you built yourself is the half that breaches you. Thirteen short lessons that turn Zero Trust from a slogan into a sequenced twelve-month build, with a crown-jewel inventory, a device tier assignment, an agent authority ledger and an evidence package you can hand an underwriter.
A distribution centre goes quiet at nine on a Tuesday morning. The conveyors stop, the scanners will not read, and upstairs the controller cannot approve a wire or reach the bank portal. Four days pass without a single box shipping. Nobody stole a database that morning. What happened was that a mid-market company lost the ability to conduct commerce, which is the part of a breach that actually ends companies, and it is the part nobody puts a number on until it happens to them.
This course puts the number on it first, then builds the architecture that contains it. The strategy is Zero Trust: assume the network is already hostile, verify every request on its own merits, and give an attacker who gets in nowhere to go. Applied to AI, it is the difference between a contained incident and a company-defining one.
What makes this a build rather than a briefing is the order. Jeff Yelton's four quarterly phases run underneath the whole course, so identity, devices and segmentation arrive in the sequence a real programme executes them, and the AI-specific work in the second half lands inside an architecture that already exists.
The real deliverable is a Zero Trust AI posture you can evidence: your crown jewels named and tiered, your devices sorted by what you can actually verify, your agents carrying a written authority ledger, your vendors isolated to one application each, and a package that answers an underwriter, an auditor and an enterprise buyer with the same documents.
CEOs and executives who are accountable when the business goes silent, whatever the technical cause.
COOs, CISOs and risk leaders who own security posture as AI spreads across the company.
CIOs, CTOs and product leaders deploying AI systems and agents that are now part of the attack surface, and that act before a human sees them.
Expected time: about 65 minutes across thirteen short lessons (four to six minutes each), plus the time you spend filling in your own artifacts.
The question that reframes every security budget conversation you will ever sit in: what does one day of downtime cost us, and can we absorb that hit three times in a quarter?
Your own breach balance sheet, with the five indirect costs that dwarf the ransom.
Zero Trust as five business assets rather than five technologies, and the reason there is no such thing as a trusted insider.
A four-phase twelve-month roadmap with a named owner per phase, and the case for why skipping the first phase costs more than doing it.
The identity control set that stops the mundane credential reuse behind most breaches, plus the three criteria to buy it on and the one exemption that collapses the whole programme.
A device tier assignment for your own fleet, built on what you can verify rather than on who owns the hardware.
A crown-jewel inventory produced by a business tabletop, and the rollout sequence that segments a network without stopping your own trucks.
The mechanism behind prompt injection and data leakage, and the red-team checks that find what standard testing misses.
An agent authority ledger: what it may do unattended, what it must escalate, what it may never do, and whose hand is on the switch.
An AI tool approval gate that channels shadow AI instead of driving it onto personal phones.
A vendor question set built on access rather than assurances, and a map from data sensitivity to deployment model.
The evidence package that serves an underwriter, an auditor and an enterprise buyer, and the five dashboard views your board can act on.
The assembled checklist, and the reason the most dangerous day is the day you declare it finished.