Shift your workforce into hyper-productivity mode with Dell Pro laptops powered by Intel® Core™ Ultra with Intel vPro®. Shop on Dell Premier.

Explore More

The FY2027 budget window closes before Thanksgiving. For most enterprise IAM teams, the next four to six weeks are the entire runway to get agent identity governance funded. After that, the number is locked, and whatever did not make it into the ask does not get its own conversation again until Q2 at the earliest, when someone has to explain why the gap was not visible six months earlier.

This issue is about winning that fight. Not the abstract argument for why agent identity matters. The argument that moves a CFO, a CISO, and Legal simultaneously, using the language each of those rooms actually responds to.

The CFO conversation is a liability conversation, not a security posture conversation. The framing that loses is "we need to improve our agent identity governance program." The framing that lands is "we have N AI agents in production with credentials that have no expiration, no ownership record, and no audit trail. If one is compromised, the exposure window is the time it takes us to notice, measured in hours. Here is what that means to the business if it surfaces in a SOC 2 audit, a breach notification obligation, or a regulatory examination." The CFO does not need to understand workload identity or SPIFFE. The CFO needs to understand that unmanaged agent credentials are contingent liability, and that contingent liability has a cost whether or not the incident happens.

The numbers that make this real are not hypothetical. The Cloud Security Alliance's 2026 NHI whitepaper found that 47% of non-human identities go unchanged for over a year, and 51% have no clear ownership on record. Your AI agents are in that population. When a CFO hears "more than half of our machine identities have no owner and have not been reviewed in twelve months," the question is not about security architecture. It is about who signed off on that.

The Legal and Compliance conversation runs on a different clock. The EU AI Act entered enforcement on August 2, 2026. Article 15 requires that high-risk AI systems maintain traceability, the ability to reconstruct what the system did, when, and on whose authority. If your agents touch EU residents, hiring decisions, credit decisions, access control, or medical records, you are in high-risk territory. An agent operating without a correlated audit trail across its tool calls does not satisfy Article 15. The Legal team does not need to understand behavioral monitoring. It needs to understand that regulators can ask for the log of what a specific agent did on a specific date, and right now that log does not exist in a form anyone can produce.

That angle closes budgets. The CFO angle creates urgency around cost. The Legal angle creates urgency around timing. The two together create a window, and the window closes when the budget does.

The CISO conversation is different because a CISO understands the technical shape of the problem. What moves a CISO is not education, it is precision. The two metrics that carry the most weight in that room are time-to-revoke and blast radius. Time-to-revoke: if a specific agent credential is compromised right now, how long before it is out of every system that accepted it? In most environments without dedicated agent identity controls, the answer is "we would have to find the credential first, which requires an inventory we do not have." Blast radius: if that credential is used during the exposure window, what is the maximum set of resources it can reach? In environments where agent credentials are scoped loosely, the answer is uncomfortable. Frame the budget ask around reducing both numbers, and you give a CISO a before/after picture that fits the risk model they already use. The exact numbers for your environment are the work you do before the meeting.

For teams starting from zero, the temptation is to wait until the program is more mature. The Q4 budget conversation is actually easier from zero, because the ask is scoped: fund the foundation. The foundation is three things. An agent inventory. An ownership model. A credential management discipline so credentials expire, rotate, and are scoped to what the agent actually does. That is a year-one program, not a moonshot. It has a cost, a timeline, and a measurable output.

What kills the ask is trying to fund everything at once. Boards and CFOs do not approve programs that cannot be described in a single paragraph. The one-paragraph version is: "We are building a governance foundation for AI agent identities, starting with inventory, ownership, and credential hygiene, delivered in four quarters, at a cost of X." Every word in that paragraph does work. None of it requires the audience to understand workload identity federation.

The meeting to avoid is the one where you show up with architecture diagrams and leave with a "we'll revisit in Q2." That happens when the ask is about technology instead of risk. The CFO, Legal, and CISO conversations above are all risk conversations. They translate agent identity posture into the language of exposure, liability, and coverage. The architecture is the answer to "how." The risk framing is the answer to "why now."

Six weeks is enough time to build the case. It is not enough time to wing it.

The budget cycle does not care that your agent identity program is technically justified. It responds to the framing that makes inaction look more expensive than action. That framing exists. The liability exposure is real, the regulatory clock is running, and the blast radius numbers are discoverable. The practitioners who get funded in Q4 are not the ones with the best programs. They are the ones who walked into the CFO's office with the right question already answered: what does it cost us to wait?

Which of those conversations is the one your organization does not yet have an answer for? Hit reply and tell me. I read every one.

See ya next week.