Operator note
Your first AI hire may be the wrong one
The reflex, when a board asks what the company is doing about AI, is to hire someone with AI in their title. Usually a senior engineer, sometimes a head of AI, occasionally a Chief AI Officer. The hire signals motion. It dates the AI strategy from the day the offer letter goes out. It also, in most B2B software companies, slows the company down.
The reflex is wrong because it misreads what is broken. The first constraint is rarely model capability or engineering bandwidth. It is the operating model the company still runs. Pricing is still per seat. Cost-to-serve still assumes humans in the loop. The roadmap still treats AI as a feature stream attached to a SaaS product. A technical hire dropped into that model accelerates the wrong work. The roadmap fills with copilots. The cost-to-serve does not move. The pricing does not move.
The board sees activity, not economics.
This is what MIT’s NANDA initiative described in August 2025, when it reported that roughly 95 percent of enterprise generative AI pilots produced no measurable P&L impact, despite 30 to 40 billion dollars of enterprise spend. The lead author named the constraint plainly: the failure was not model quality, it was the gap between the tool and the workflow it was supposed to enter (Fortune / MIT NANDA, August 2025). McKinsey’s State of AI, March 2025, ran the same finding from a different angle. Of 25 organisational attributes tested across 1,491 respondents, workflow redesign had the single largest correlation with EBIT impact from generative AI. More than 80 percent of organisations using generative AI reported no tangible enterprise-level EBIT effect (McKinsey, March 2025). BCG’s Build for the Future 2025, published September 2025, found only 5 percent of 1,250 companies were generating real returns. The 60 percent in the laggard tier were not behind on technology. They were behind on operating model (BCG, September 2025).
The false diagnosis is to read this as a talent problem. The boardroom logic runs as follows: we have an AI gap, therefore we need an AI leader, therefore we hire. IBM’s 2026 CEO study reports that 76 percent of organisations now have a Chief AI Officer, up from 26 percent a year earlier (IBM Institute for Business Value, May 2026). The role is being filled at speed. The role is also, in many companies, being defined narrowly. One European survey of 2025 CAIO appointments found two-thirds of incumbents spent their first twelve months building an AI inventory and documenting use cases, not setting strategic direction. Harvard Business Review made the structural point in August 2025: appointing a single AI leader to a vaguely scoped mandate frequently fails, because the role is too broad and too disconnected from where business decisions actually get made (HBR, August 2025).
The reason it fails is mechanical. The economic shift AI forces in B2B software is not at the feature layer. It is at the pricing layer, the cost-to-serve layer, the work-definition layer, and the control point that determines where value is captured. Tomasz Tunguz has documented this shift in 2024 and 2025: when AI agents do work that previously required seats, seat-based pricing stops describing the value, and software begins to be priced closer to outcomes or to compute, with gross margin patterns that look more like infrastructure than SaaS (Tunguz, 2024 and 2025). A technical lead, however strong, cannot move pricing, cost-to-serve, or work definition without authority over them. Those decisions sit with the CEO, the CFO, the CRO, and the CPO. A first AI hire who reports below that line will, by structure, end up building AI features inside the existing operating model. That is the failure mode named in the MIT, McKinsey, and BCG data.
The intervention is to rethink the operating model first, and only then sequence the technical hires. Concretely, this means three pieces of work that precede the job spec.
The first is a margin and pricing diagnostic. Where will gross margin compress as AI raises customer expectations of the work? Where can pricing move from access to outcome, and what does that do to renewal quality and net retention? McKinsey’s State of AI found workflow redesign was the largest single lever on EBIT impact, and BCG’s leaders captured 1.7 times the revenue growth and 1.6 times the EBIT margin of laggards. The numbers describe a pricing and margin reset, not a feature backlog.
The second is a work-definition map. What is the unit of work the product is paid for today, and what unit will it be paid for in two years? If the answer is the same, the AI roadmap is decorative. If the answer is different, every downstream decision, sales motion, customer success staffing, support model, and packaging, must change before scaled engineering hires.
The third is the leadership architecture. HBR’s August 2025 framing is correct on this point: AI accountability cannot sit in one role. It has to be distributed across the CEO, CFO, CRO, and CPO, with a strategist or transformation lead who can hold the operating model question against pressure from product to ship more features.
Only after these three pieces are in place does the first technical hire pay back. At that point, the engineer or applied scientist is solving a defined economic problem, not assembling a feature list.
What changed and what did not. The technical hire is still needed. The sequence in which it is made now determines its return. The board question worth replacing is “do we have a head of AI yet.” The better one is “what has the operating model decided to become, and what hire does that decision now demand.”
Rule of thumb. Before signing the first AI job offer, write the one-page memo that names the pricing unit, the cost-to-serve target, and the work the product will be paid for in two years. If that page cannot be written, the hire is premature. The next hire after that page exists is usually a strategist or operating lead, not an engineer. The engineer comes third, into a brief that is already economic.
Sources
- The state of AI: How organizations are rewiring to capture value, McKinsey, March 12, 2025.
- MIT report: 95% of generative AI pilots at companies are failing, Fortune (reporting on MIT NANDA, The GenAI Divide: State of AI in Business 2025), August 18, 2025.
- Are You Generating Value from AI? The Widening Gap (Build for the Future 2025), Boston Consulting Group, September 2025.
- IBM Study: CEOs are Reshaping C-suite Roles for the AI Era, IBM Institute for Business Value, May 4, 2026.
- Your AI Strategy Needs More Than a Single Leader, Harvard Business Review, August 4, 2025.
- No SaaS! How AI Agents Will Change Software Pricing, Tomasz Tunguz, 2024.
- The Unsustainable Subsidy, Tomasz Tunguz, 2025.
- Chief AI Officer 2026: Real Role or Just Another C-Level Title?, Digital Chiefs, 2026.