How to Use This Worksheet

This worksheet operationalizes Step 4 – Phase 3 and Step 7 of the AI Budget Governance Playbook. Work through it jointly as CFO, CTO, and AI leader, using the AI portfolio, scenario bands, and GenAI cost drivers you already modeled in Phase 2.

Recommended sequence:

1. 01Confirm scope and constraints of your AI budget envelope.
2. 02Define AI spend guardrails, usage caps, and quotas per risk category.
3. 03Attach guardrails to budget lines, business outcomes, and governance rhythms.
4. 04Specify risk contingencies for usage spikes, vendor/model switches, pricing shifts, and vendor lock-in.
5. 05Overlay sector- and regulation-specific adjustments and controls.
6. 06Summarize decision rules into a spreadsheet or planning tool for ongoing AI FinOps.

Section 1 – Confirm AI Budget Envelope and Scope

Use this section to restate the AI budget envelope in a way that is actionable for guardrail design.

1.1 AI Budget Envelope Summary

Fill in with current-year values (extend for forecast years as needed).

Prompts:

• Which budget lines are usage-based vs. fixed/committed?
• Which lines are driven primarily by tokens, which by compute, and which by licenses?

1.2 Scope of Guardrails

Clarify where AI spend guardrails, usage caps, and quotas will apply.

• Guardrails apply to:
• [ ] All GenAI and LLM spend
• [ ] Only usage-based GenAI spend
• [ ] Only initiatives in certain risk categories (specify):
• Exclusions (if any) and rationale:
• Link to AI portfolio artifact (location / owner):

Section 2 – Guardrail Design by Risk Category (Step 4)

Work risk-category by risk-category as defined in the playbook (e.g., Low, Medium, High, Critical). The goal is to standardize guardrails while allowing targeted flexibility.

2.1 Risk Category Guardrail Template

Copy this subsection once per risk category.

Risk category name:

Typical use cases in this category:

• Example initiatives (reference AI portfolio IDs):
• Primary business outcomes targeted (e.g., revenue, cost, risk reduction):

A. Spend Guardrails, Caps, and Quotas

Use this table to define the high-level posture for this risk category.

B. Approval Thresholds and Decision Rights

Define who must be involved when spend or usage trends toward or exceeds guardrails.

Section 3 – Guardrails Linked to Budget Lines and Outcomes

This section connects each AI spend guardrail to specific budget lines, business outcomes, and governance rhythms, as required by the playbook.

3.1 Budget Line Guardrail Register

Use this as the master table; maintain a version-controlled copy in your AI budget governance spreadsheet.

Instructions:

1. 01For each budget line from Phase 2, assign a risk category and primary business outcome (as per Step 6 of the playbook).
2. 02Specify at least one AI spend guardrail and, where usage-based, both a usage cap and a quota.
3. 03Tie each line to a scenario band (e.g., Base, Aggressive, Downside) from your GenAI cost driver model.
4. 04Define the governance rhythm where adherence is checked (e.g., monthly AI FinOps huddle, quarterly portfolio review, board update).

Section 4 – Risk Contingencies and Decision Rules

This section makes risk contingencies explicit for:

• Usage spikes
• Model or vendor switches
• Pricing shifts
• Vendor lock-in risk

4.1 Usage Spike Contingencies

Define structured responses when usage sharply exceeds plan.

Checklist – Usage Spike Policy

• [ ] Define what constitutes a "spike" (e.g., >30% above quota for more than 7 days).
• [ ] Set per-risk-category spike thresholds and actions.
• [ ] Ensure monitoring dashboards show near real-time usage and forecast.
• [ ] Pre-define communication templates to business owners when throttling.
• [ ] Specify override process and who can authorize temporary cap increases.

Usage Spike Decision Table

4.2 Model or Vendor Switch Contingencies

Capture rules for switching models or vendors, especially when responding to cost, performance, or risk.

4.3 Pricing Shift Contingencies

Specify rules to respond to vendor pricing changes while staying within the AI budget envelope.

Pricing Shift Rule-of-Thumbs

• Mild pricing change (±10%): handle within existing budget lines; adjust quotas but keep guardrails.
• Moderate change (±10–30%): re-run scenario bands; propose adjustments in next quarterly AI FinOps review.
• Severe change (>30%): trigger re-opening of the AI budget envelope and potentially board-level review for material impact.

4.4 Vendor Lock-in Risk Contingencies

Define how you manage vendor concentration risk and mitigate lock-in.

Section 5 – Sector- and Regulation-Specific Adjustments (Step 7)

Use this section to overlay sector/regulatory constraints onto the guardrails above. Do this jointly with Legal, Compliance, and Security.

5.1 Sector & Regulatory Profile Snapshot

• Sector(s) (e.g., banking, healthcare, public sector):
• Jurisdictions (e.g., EU, US, UK, others):
• Applicable AI or data regulations (e.g., EU AI Act risk level, HIPAA, GLBA, GDPR, sector-specific guidelines):
• Internal risk appetite statement (link or summary):

5.2 Regulatory-Driven Guardrail Adjustments

Map where generic guardrails must be tightened (or cannot be relaxed) due to regulation.

5.3 Sensitive Use Cases and Shadow AI Controls

Identify use cases that require elevated controls, including shadow AI costs and unsanctioned tools.

Checklist – Sensitive Use Case Controls

• [ ] Catalog all use cases involving regulated data (PHI, PII, payment data, trade secrets).
• [ ] Assign them to High or Critical risk categories in the AI portfolio.
• [ ] Apply stricter usage caps, quotas, and monitoring frequency.
• [ ] Prohibit unsanctioned GenAI tools for these data categories.
• [ ] Allocate specific budget for monitoring and discovery of shadow AI costs.

Section 6 – Governance Rhythm Integration

Confirm where and how these guardrails, caps, quotas, and contingencies are monitored and adjusted.

6.1 Governance Rhythm Summary

6.2 Owner and Review Cadence

• Primary owner of this worksheet:
• Last updated:
• Next scheduled review:

Section 7 – Final Consistency Checks

Before finalizing, run this quick checklist to ensure coherence with the AI Budget Governance Playbook.

Checklist – Consistency with Playbook Artifacts

• [ ] Every AI initiative in the AI portfolio has a mapped budget line, risk category, and business outcome.
• [ ] Each budget line has at least one clearly defined AI spend guardrail and, where usage-based, explicit usage caps and quotas.
• [ ] All guardrails and risk contingencies reference the same scenario bands used in your GenAI cost driver model.
• [ ] Risk contingencies exist for all four areas: usage spikes, model/vendor switches, pricing shifts, and vendor lock-in.
• [ ] Sector- and regulation-specific constraints have been applied to High/Critical risk categories and sensitive data use cases.
• [ ] Governance rhythms (monthly AI FinOps, quarterly reviews, board updates) clearly indicate where guardrails and contingencies are monitored and adjusted.

Once this worksheet is complete and aligned with your spreadsheet model and intake forms, you have effectively executed Step 4 – Phase 3 and Step 7 of the playbook and can move on to integrating these decisions into your ongoing AI FinOps and budgeting cycles.