Revenue Strategy
The Revenue AI Gap
Why the teams with the most to gain from AI are capturing the least of it
McKinsey estimates that generative AI could unlock up to $1.2 trillion in productivity for sales and marketing, more than any other function in the enterprise. And yet, by virtually every measure that matters, revenue organizations are among the slowest to claim it.
That is the central paradox in enterprise AI. According to Worklytics and Larridin's 2026 AI benchmarks, engineering teams at mature companies are running at 80 to 90 percent meaningful AI adoption. Sales, marketing, and customer success at the same companies, with the same leadership and the same budgets, are operating at 20 to 40 percent. Same company. Four times the adoption rate.
Engineering
80–90%
Sales, marketing and customer success
20–40%
Source: Worklytics and Larridin, 2026 AI benchmarks
The default explanation is cultural resistance. I do not think that fully explains the issue. After sitting with more than a dozen CROs and CEOs over the past year, I am convinced the issue is not unwillingness. It is structural. Three organizational failures, operating simultaneously, make meaningful adoption a persistent uphill battle even for willing teams.
The fluency problem runs from the top down
The first wave of AI tools was built for content creation and information retrieval, which is exactly the work senior leaders have largely delegated. But the deeper issue is that engineers have spent years building with these tools, breaking them, integrating them into real personal and professional workflows, and developing genuine intuition for where they help and where they fail. Most GTM professionals have not. They were late to the party, and their experimentation has been limited to basic tasks like email generation. That is not a recipe for fluency. It is a recipe for avoidance.
Context debt is the hidden tax on every AI interaction
Every morning, a sales rep opens a chat window and re-explains the company's ICP, competitive positioning, and quarterly priorities just to get a usable first draft. When the session ends, the context evaporates. The next rep starts from zero. So does the same rep the following morning. The AI is like a perpetual intern who shows up every Monday with no memory of last week.
This creates context debt, and it compounds. It explains why so many organizations report genuine time savings from AI while seeing no movement in win rates or pipeline quality. The activity metrics improve. The outcome metrics do not.
The activity metrics improve. The outcome metrics do not.
Distrust is rational, not irrational
Varicent's 2025 research found that 44 percent of revenue leaders cite their own skepticism as a greater barrier to AI value than any technical issue. That is often viewed as a mindset problem. I read it differently. A hallucinated detail in a client brief can derail a six-figure deal. A miscalibrated forecast triggers the wrong resource decisions upstream. The sensitivity to AI error in revenue is not conservatism. It is appropriate risk calibration.
The problem deepens when AI becomes a substitute for thinking rather than an accelerant for it. Reps who consistently outsource their analytical work stop developing the critical thinking that made them effective in the first place, including pattern recognition, objection handling, and deal signal instincts. When AI does that work, the rep gets the output without building the skill. That is not a technology failure. It is a discipline failure, and it compounds quietly until the damage is already done.
A framework for breaking the loop
My clients who have made the most meaningful progress closing this gap follow a three-phase approach: Crawl, Walk, Run.
01
Crawl
Eliminate context debt
A shared knowledge base with the ICP, competitive positioning and objection library that AI draws from automatically.
02
Walk
Standardize the workflows
AI-assisted routines for high-leverage, repeatable tasks: pre-call research, account scoring, post-call follow-up, churn flags.
03
Run
Connect agentic systems
Multi-step workflows tied to systems of record that update deal stage and forecast category and surface deal health signals.
Crawl eliminates context debt. The output is a shared institutional knowledge base, a living repository containing the organization's ICP, competitive positioning, and objection library that AI draws from automatically rather than requiring reps to reconstruct it from scratch each session. After building this system, one of my clients reduced manual meeting prep time by roughly 60 percent. The bigger win was that bottom-half AEs started showing up to calls as well-prepared as top-half AEs. AI did not raise the ceiling. It raised the floor.
AI did not raise the ceiling. It raised the floor.
Walk standardizes AI-assisted workflows for high-leverage repeatable tasks: pre-call research, account scoring, post-call follow-up, churn flags. Not glamorous work. But in organizations where the top 17 percent of reps generate 81 percent of revenue, as found by Ebsta and Pavilion, raising the floor has a massive impact on productivity.
Run is where agentic applications enter: multi-step workflows that connect the shared brain and skill layer to your systems of record. A well-designed Run-phase system might ingest call transcripts, update deal stage and forecast category automatically, and surface deal health signals to the rep before they think to look. Teams may employ a live client ‘stock ticker’ that continuously surfaces account health to key stakeholders. Published case studies, including from LangChain, at this level show 250 percent increases in conversion rates. But they are built on top of functioning Crawl and Walk infrastructure. A word of caution: connecting the systems is the easy part. The real work is in the data architecture underneath, and most organizations underestimate it significantly.
The CRO as Architect
The traditional CRO archetype is a motivator, a talent developer, a pipeline quarterback. That job description is not obsolete. But the challenge described here is not a people-management problem. It is an information architecture problem.
The CRO who succeeds in this environment is the architect of the shared brain. Their most consequential decisions are no longer about headcount, territory design, or commission structure. Those are table stakes. Instead, they are about what context their team operates from, what workflows they run, and whether the system they are building is measurably smarter every quarter than it was the quarter before.
That shift, from managing people to architecting the environment in which people and AI work together, is what separates the next generation of revenue leaders from the last one.

