AI Strategy · Budget Optimisation

The AI Portfolio
Audit Framework

A structured 5-step process to eliminate wasted AI spend, identify what's actually production-ready, and redirect budget toward initiatives with measurable P&L impact.

95% of GenAI pilots produce
zero P&L impact in year 1
42% of companies scrapped
AI initiatives in 2025
40% avg. AI budget saved
within 30 days
Get the full framework

Enter your details to unlock all 5 steps and download a PDF copy.

  • How to inventory every active AI initiative
  • The P&L Test that kills vanity projects instantly
  • Kill / Accelerate / Wait classification template
  • Production readiness checklist for each blocker type
  • Budget reallocation governance framework
Please enter your name and a valid email.

No spam. Unsubscribe anytime.

✓
Framework unlocked

Scroll down to read all 5 steps, then use the Download PDF button to save your copy.

The 5-Step Process
1
Discovery · Week 1
Inventory — Map Every Initiative
Build a single register of every active AI project: tool, vendor, monthly cost, initiative owner, current status, and what success was supposed to look like when it was approved. Most organisations discover they have 30–50% more running than leadership believes. Shadow AI spend is the norm, not the exception — licence fees quietly renewing, PoC environments still live, vendor pilots that never formally closed.
2
Filter · Week 1
P&L Test — Revenue, Cost, or Risk?
For each initiative, ask one question: does this demonstrably touch revenue, reduce cost, or mitigate a quantifiable risk? Answers like "improves employee experience" or "builds AI culture" fail the test. These may be real benefits but they are not P&L items. Flag anything that cannot be tied to a hard business metric — it belongs in a separate innovation budget with explicit, time-boxed tolerance for ambiguity.
3
Decision · Week 2
Kill / Accelerate / Wait
Classify every initiative into one of three buckets. Kill — no credible path to production or ROI within 6 months; terminate contracts and reassign resource. Accelerate — clear business case, specific blocker preventing production; this gets dedicated attention in Step 4. Wait — viable but lower priority or dependencies unresolved; park with a defined review date. This single step typically frees 35–45% of active AI spend.
4
Unblocking · Week 2–3
Production Readiness Check
For each Accelerate initiative, diagnose the specific blocker preventing it from going live. Blockers fall into five categories: data access (pipelines not connected, quality issues), system integration (API gaps, security review pending), compliance sign-off (legal or risk not engaged), unclear ownership (no named accountable person), or missing evaluation criteria (no agreed definition of done). Name one blocker, one owner, one resolution date — and hold them to it.
5
Outcome · Week 3–4
Budget Reallocation
Redirect freed budget exclusively to Accelerate initiatives or hold it in reserve until a new initiative passes the P&L Test. Equally important: establish a lightweight governance gate so this pattern doesn't repeat. Any new AI spend above a defined threshold must answer the P&L Test, name a production owner, and set a go-live date before approval. The audit itself is a one-off; the gate makes it permanent.
Expected Outcome — 30 Days

Organisations that complete this audit typically cut active AI spend by 30–40% while simultaneously accelerating their 2–3 highest-value initiatives to production. The result isn't less AI — it's AI that actually ships and shows up in the P&L.

Jayashankar Attupurathu — AI Strategy & Budget Optimisation · attupurathu.in