The 7 Questions Every Operating Partner Should Ask About Portco AI

In the regular business review with your portfolio company's CEO and CFO, you will hear about AI. Maybe they've deployed a claims processing agent. Maybe they're piloting an AI sales assistant. Maybe they have an "AI strategy" that involves studying ChatGPT. But in most cases, you will not have the operational language to push back on their answers. This article provides seven questions that every operating partner should ask in every quarterly or semi-annual AI deep dive. These questions are designed to surface real value creation (or real problems) and move the conversation from strategic posturing to operational reality.

Question 1: What's Our Cost-Per-Outcome Trend Across Portcos?

This is the opening question. It establishes whether your portfolio companies are even measuring AI efficiency, and whether you have portfolio-wide visibility.

The answer you're looking for: "For our claims operation, cost per adjudicated claim is $28.50, and it's trending down 2.4% quarter over quarter. For our customer service operation, cost per resolved ticket is $3.82 and stable. We track this monthly across all AI-assisted workflows."

The red-flag answers:

• "We don't have a single cost-per-outcome KPI; each operation measures success differently."
• "Cost per claim is roughly $30, but we're not sure about the breakdown."
• "We're focused on ROI, not cost per unit."

Why this matters: Cost per outcome is the operating partner's true insight into whether the portfolio company has built sustainable, measurable AI value creation. If they're tracking it, they understand the economics. If they're not, they are operating blind. And you cannot compare performance across portcos without this metric.

The follow-up: "Can you show me the trend line for the last 6 months? Is it stable, improving, or degrading?" Stable is fine. Improving is great. Degrading is a warning sign—either the AI system is faltering, or humans are having to review more work due to accuracy issues.

Question 2: Which Portcos Can't Tie an AI Dollar to a Specific Business Outcome?

This is your shadow AI detection question. You're asking: where in your portfolio is AI spend invisible or unaccounted for?

The answer you're looking for: "We've audited all portcos. Three of them have AI spend that's clearly allocated to specific workflows. Two of them have pockets of untracked spend in cloud infrastructure and contractor invoices—we're consolidating that now. One small portco has no AI spend yet."

The red-flag answers:

• "Most of our portcos have some AI spend we can't fully trace."
• "We know there's AI spend in cloud bills, but we don't break it out separately."
• "One of our bigger portcos says they're using AI, but we're not sure where."

Why this matters: If you cannot tie AI spend to outcome, you cannot make capital allocation decisions. You might be investing heavily in a portco's AI initiative that is not actually driving value. Or you might be leaving money on the table in a portco where AI is working but underfunded. Visibility is prerequisite to governance.

The follow-up: "What's your timeline for getting all portcos to stage 3 of the maturity curve [allocated AI spend by business unit]?" The answer should be "Q3" or "Q4"—not "we'll get to it eventually."

Question 3: What's Our Vendor Concentration Risk?

This is your de-risking question. You're asking whether your portfolio is dependent on a single vendor or if you have diversification.

The answer you're looking for: "OpenAI accounts for 62% of our portfolio AI API spend, Anthropic 28%, Google 7%, and specialized vendors for healthcare compliance and document processing 3%. We've negotiated volume discounts with OpenAI and have SLAs in place. If OpenAI went down, we could shift to Anthropic within 72 hours for most workflows."

The red-flag answers:

• "We're heavily dependent on OpenAI, but we don't have a backup plan."
• "We're not sure which vendors we're using across the portfolio."
• "One of our portcos is built on a proprietary vendor, and there's high switching cost."

Why this matters: Vendor concentration is a real financial risk. If OpenAI's pricing rises 30%, or if Anthropic has an outage, your portfolio's economics are at risk. Diversification across vendors (and across commodity APIs versus custom models) is a basic risk-management discipline, like not concentrating your supplier base in one geography.

The follow-up: "If OpenAI raised prices 20% tomorrow, which portcos would feel the most pain, and what would we do?" A good CFO can answer this in a sentence. If they cannot, you have a problem.

Question 4: Where On the 5-Stage AI Cost Maturity Curve Is Each Portco?

This is your maturity benchmark question. It tells you which portcos are operationally organized around AI economics and which are still figuring it out.

The answer you're looking for: "Our healthcare claims operation is at stage 4 [AI spend tied to work items with cost-per-outcome KPI]. Our insurance underwriting is at stage 3 [AI allocated by business unit]. Our staffing operation is at stage 2 [AI spend tracked but not broken down]. Our hospitality operation has no AI yet. By year-end, we're targeting stage 3 or above for all AI-active portcos."

The red-flag answers:

• "I'm not familiar with that maturity model."
• "Most of our portcos are probably at stage 1 or 2, but I haven't assessed them."
• "We don't really measure maturity; we just deploy and optimize as we go."

Why this matters: The maturity curve is a diagnostic tool. A portco at stage 1 (invisible spend) is at higher risk of waste and lower value creation. A portco at stage 4 (cost per outcome tracked) has the operational discipline you want. Knowing where each portco sits tells you where to focus attention and investment.

The follow-up: "What does it take to move a portco from stage 2 to stage 3?" The answer should involve finance discipline, not technology. It's about setting up data flows and accountability, not building new systems.

Question 5: What's Our Portfolio-Wide AI Gross Margin Trajectory?

This is your financial impact question. You're asking whether AI is actually improving the bottom line.

The answer you're looking for: "Last year, AI initiatives across the portfolio contributed to 180 basis points of gross margin expansion—about $3.8M on our $210M portfolio revenue. We're tracking this quarterly and have visibility into which portcos are driving the most margin."

The red-flag answers:

• "We're not calculating portfolio-wide AI margin impact yet."
• "We know individual portcos have AI initiatives, but we haven't modeled the aggregate effect."
• "It's too hard to isolate the AI impact from other operational improvements."

Why this matters: This is the partner conversation. If you cannot articulate the gross margin impact of AI across your portfolio, you cannot speak to the board about value creation. And if AI is not improving margins, you need to ask why you're investing in it.

The follow-up: "Which portcos are over-indexing on AI margin expansion, and which are lagging? What's your plan for the laggards?" This reveals whether your portfolio has a coherent AI playbook or if different portcos are making independent bets.

Question 6: Which Portcos Have Shadow AI We Can't See?

This is your discovery question. You're asking whether there's unaccounted AI spend hiding in business-unit budgets, contractor invoices, or personal credit cards.

The answer you're looking for: "We did a shadow AI audit last quarter. We found $340K per year of hidden spend: cloud infrastructure that wasn't being tracked, a contractor running training jobs on AWS, and subscription licenses buried in office overhead. We've consolidated it and brought it into the portfolio cost system. Going forward, all AI spend over $5K per month requires CFO approval."

The red-flag answers:

• "We haven't done a shadow AI audit."
• "I'm sure there's some spend we're not seeing, but we don't know how much."
• "One portco mentioned they have some AI pilots, but we're not tracking the spend."

Why this matters: Shadow AI is the tax you pay for lack of governance. The FTI study found 40% of PE portfolio AI spend invisible to the chief investment office. That is your gap. If you cannot see it, you cannot govern it, optimize it, or allocate capital to it.

The follow-up: "When do you plan to do your next shadow AI audit?" The answer should be "every quarter as a routine control." If it's "we'll get to it eventually," that's not a control.

Question 7: What's Our AI Exit Story for the Next 18 Months?

This is your forward-looking question. It connects AI to the actual business outcomes that matter: exit readiness and valuation.

The answer you're looking for: "For our healthcare claims portco, which we're likely exiting in 18 months, the AI story is: we've built a sustainable claims adjudication operation that costs $28 per claim (down from $42 manual), with cost attribution proving the efficiency to buyers. We've documented the workflows, trained a second team member as backup, and have no vendor lock-in. The buyer will inherit a mature operation. For our staffing portco, we're still in stage 2 and probably need another holding period; AI is not yet core to the exit story."

The red-flag answers:

• "We're not thinking about the exit story yet."
• "We'll figure out the AI narrative when diligence starts."
• "We're not sure which portcos will have a strong AI story by exit."

Why this matters: You are an investor. Your job is to exit successfully. If AI is material to a portco's value, the exit story must be bulletproof. If you are three months from exit and the buyer's advisors discover hidden AI spend or unsustainable cost models, you have a negotiating problem. Thinking about the exit story 18 months out gives you time to fix it.

The follow-up: "For each portco you're planning to exit in the next 24 months, can you walk me through the AI value-creation narrative and the cost attribution proof?" If the CFO cannot answer this in a coherent paragraph per portco, that is work to be done before exit.

---

These seven questions move AI from strategic conversation to operational accountability. A CFO or COO who can answer all seven clearly has built disciplined AI governance. One who cannot has visibility or governance gaps worth addressing. Use these questions in your board meetings, your quarterly business reviews, and your exit planning. They are the operating partner's guide to real AI value creation.

For deeper guidance on implementing AI governance frameworks, cost attribution systems, and exit-readiness narratives across your portfolio, see the PE Operating Partner AI Playbook. Operating partners running this analysis across a portfolio can request the PE Operating Partner Field Guide with templates for tracking these seven KPIs across all portcos.