Your AI Pilots Won't Move the P&L. Here's What Will - Adam Nahari, Apollo
Key Takeaway
Moving the P&L with AI requires investing five units of change management and end-to-end workflow redesign for every one unit of engineering.
Highlights
Nearly 90% of businesses report using AI in at least one business function, but only 6% report significant EBIT impact.
Generative AI time savings amount to about 2% of total US working hours, yet almost none of that shows up in the P&L.
For every five units of change management in an AI transformation, there is only one unit of engineering.
Businesses that see material EBIT impact are three times as likely to redesign an entire workflow end to end rather than simply adopting a new tool.
An optimization agent engine built for airline crew scheduling reduced solution generation time from 15 minutes to under a minute.
Projects fail or succeed based on explicit kill metrics, including 90-day time gates and an adoption threshold requiring over 50% usage by schedulers.
Timeline
The AI Productivity Paradox
Manual crew scheduling often results in no legal solutions, forcing the airline to rely on expensive premium pay. Despite high adoption rates across 90% of businesses, only 6% experience significant EBIT impact. Technical engineering represents only a fraction of the overall transformation effort, while selecting problems, building buy-in, and driving adoption dictate overall success.
- Cargo airline flight schedulers face 15-minute windows to resolve crew assignments manually using rest rules and qualifications.
- Over 55% of US adults have used generative AI within three years, outpacing adoption curves of the PC and internet.
- Change management consumes five times more effort than engineering during successful AI transformations.
Three Layers of Business AI
Businesses divide AI investments across three distinct layers ranging from generic copilots to custom core operational systems. Because core operational tools cannot be purchased off-the-shelf, their financial gains remain protected in profit margins. High-performing companies heavily favor end-to-end workflow redesign over simple tool adoption.
- General productivity tools like copilots and chatbots show gains in price rather than margin because competitors share access.
- Domain-specific AI tools leverage retrieval over enterprise knowledge to improve internal business function efficiency.
- Core operational AI embeds vertical systems directly into foundational business processes to protect margins.
The First Three Moves for Transformation
Targeting primary cost drivers like airline crew scheduling focuses capital on metrics recognized directly by the CFO. Replacing manual scheduling with an optimization agent engine turns a marginal time-saving play into a 10x operational upgrade. Without direct public sponsorship from the executive suite, initiatives encounter persistent scheduling conflicts and delayed data access.
- Transformations start by targeting one or two major financial drivers rather than spreading resources across 40 different pilots.
- Redesigning a process entirely around new technology yields transformational gains compared to incremental gains from basic automation.
- Public commitment from the CEO serves as a strict prerequisite for funding and cross-departmental alignment.
Execution, Tooling, and Governance
Recruiting specialized leadership with combined technical and domain expertise bridges the gap between engineering teams and end users. Establishing rigorous kill gates and working alongside the CFO to verify compute and token scalability prevents unprofitable scaling. Optimization agents ultimately compress complex 15-minute manual coordination tasks into workflows taking less than one minute.
- Building solutions alongside business co-sponsors and embedded AI champions ensures high user adoption.
- Executive leadership cannot effectively redesign workflows with AI without personal experimentation and daily tool usage.
- Strict kill metrics, 90-day time gates, and unit economic evaluations determine whether pilot projects survive.
Description
Most companies use AI, but very few see it show up in their margins. Adam Nahari of Apollo Global Management explains why engineering is the smaller part of an AI transformation. He walks through the six moves that turned an airline's crew-scheduling desk into a 10x win: start from the P&L, pick one or two problems, get a real owner, put the tools in everyone's hands, and set kill metrics before you start. 0:00 Intro: the airline crew scheduling desk 1:22 AI adoption vs. real EBIT impact 2:42 Why engineering is the smaller part 3:42 The three layers of AI value 5:45 Start from the P&L 7:21 The 10% play vs. the 10x play 9:02 Getting it used: owners and alignment 11:42 Put the tools in everyone's hands 12:40 Set kill metrics first 14:07 Recap: the six moves Adam Nahari, Apollo Global Management: https://www.apollo.com Join us in SF for the second annual AIE CODE summit: https://aie.codes our last flagship summit of 2026. Applications live now!