Ten Questions for your Next Board Meeting
Every portfolio company now has an AI slide and a set of chatbot/co-pilot/agents. Very few have an AI strategy a PE or Strategic buyer will underwrite.
With holds running long and buyers pricing AI durability into every software deal, operating partners need a faster way to tell the difference. These are the ten questions I would ask a portfolio CEO and product leader before the next board meeting.
Strategy and outcomes
Which customer outcome does our AI move, and by how much? Not features shipped. A measurable outcome: cycle time, error rate, cost to serve, retention. If the answer is a list of features, the team is running a feature factory with a new label.
Is AI in the core workflow or next to it? A slide out chatbot panel that answers questions is table stakes. The question is whether the product's primary job gets done differently.
What happens to revenue if our customers' seat counts halve? If pricing is tied to human users and the product succeeds at automation, it cannibalizes itself. Model it before a buyer does.
Architecture and trust
Which parts of the product are deterministic, and which are probabilistic? The team should be able to draw that line clearly. If the AI is doing the math that must be right, that is a risk.
Can a user see where every AI-touched output came from? Glass box by design, not an activity log after the fact.
Do we reason over business meaning or customer-specific configuration? A semantic layer is the difference between an AI that scales across customers vs one that is re-implemented for each.
Can our customers' agents or LLMs call our product? If services are not exposed through machine-readable contracts, the product risks becoming a screen other agents work around.
Where is the autonomy boundary? Human approval should be a measured starting point, with acceptance and override rates driving expansion.
Operating model
How has the way we build software changed? If the product and engineering organization has not adopted agentic development in its own lifecycle, it will struggle to ship agentic products at the pace the market now expects.
Who owns the outcome? Empowered product teams accountable for outcomes, or a roadmap of committed features handed down from the executive team? The AI era does not change this principle of the product operating model. It raises the cost of ignoring it.
Key takeaways
Most portfolio companies have an AI slide, not an AI strategy a buyer will underwrite.
Ten questions across strategy, architecture, and operating model separate the two.
The pattern of answers tells you whether the fix is strategy, architecture, or the operating model, and how much of the hold you need.
