The AI Value Trap.
Why the operating model, not the technology, decides whether AI value
reaches the P&L.
Anyone can build now. Who owns what gets built?
AI has democratised access to the tools that deliver change. Anyone in a company can describe what they want in plain language and have a model write the software, so engineering capacity is no longer the limit.
Across the PE-backed and mid-market companies we work with, the same pattern keeps appearing. The tools work and people are building, yet the effect on the P&L is hard to find. The paper argues these are symptoms of one gap: there is no operating model for owning what AI makes easy to build.
When anyone can build a working capability in an afternoon, what controls the change mechanism?
- Who owns what gets builtTools that nobody owns do not get integrated, trusted or used.
- How the flow is redesignedAutomating a single task inside an obsolete process leaves every handoff, approval and exception intact.
- Whether people trust itA tool nobody trusts does not get used, and an unused tool never reaches the P&L.
Three places value erodes.
From an investor’s perspective, the thesis begins to fail in three key areas.
Headcount out, model unchanged
Headcount is removed from the plan to match an AI business case, but the operating model around the work stays the same. The savings are absorbed by workarounds, overtime and re-work.
Integrations that break
Integrations built outside any standard break without notice, sometimes damaging data in core systems such as the ERP. Vibe coded integrations are especially prone to this.
Data leaving the perimeter
Operational data is sent to external model endpoints with no check on residency or sovereignty, and no one has defined what happens when an agent gets a calculation wrong.
Separate creation from control.
None of this is an argument against democratisation or against vibe coding. Sending teams back into a months-long ticket queue would stop the creativity that produces value. The paper sets out nine actions.
Get the full paper.
The detail behind the brief.
- Why this differs from earlier shadow IT, and the two risks to manage
- How islands and orphans show up in buyer diligence
- All nine actions, with how to set up a control desk
- The conclusion: three things that decide whether building compounds into value or risk
