We've talked to 100+ companies about internal AI adoption, and through our research we've found that what lets a company improve on AI is less about how skilled individual employees are with AI and more about how the company manages AI ownership.
Engineering built a system; go-to-market stopped at tools
Late 2022 was the ChatGPT moment. Every company adopted it the same way: a few people tried it, then everyone did.
Engineers picked their AI tools and then built a system around them, because that is what engineers do: scaffolded data, centralized tooling, standard metrics. Best practice spread on its own, even if nobody planned how it turned out. Engineering is now a fundamentally different job, one that looks little like the one it replaced.
Go-to-market picked its tools and stopped there. The technology was too big a bite for any one function to take on its own, and no other function came with engineering's habits. A few citizen builders got very good at AI… but the rest of the organization never transformed.
4 years on, AI in the business has no numbers and no owner
The promise of AI – better efficiency, faster pipeline, more revenue – still feels like science fiction. Nobody can measure AI's business impact, though we all plod ahead anyway.
Two teams often step up to help:
- IT or RevOps takes the tools: roadmap, vendor evaluation, and deployment.
- Enablement, often inside HR, helps humans use AI better.
Everyone's trying.
Despite best efforts, AI adoption outside of engineering is a slog and it's not clear where we're heading. Teams ask for more tools and more training. Vendors deliver more work for RevOps.
The skills spreadsheet is six months stale. Citizen builders do their thing, and the rest attend the lunch and learn but their job doesn't meaningfully change. Meanwhile, the CEO isn't going to stop asking about those "AI efficiency gains" any time soon.
Sometimes, the slog has a cause: nobody owns the change
RevOps owns the tools and Enablement owns the training, but neither can change how a sales rep does their job. Engineering's answer is that everyone should become an engineer, which does not work for sales, marketing, or customer success.
Without an owner senior enough, bold enough, and technical enough to change how people work, nothing moves. But the right person, armed with the right data, can lead a team through the discomfort of a generational change in the way work gets done.
Run AI the way you run the business, with data
Leaders, we see you. There is a better way, and it's the one we know and love when we build our businesses: data-driven, outcome-focused, and human-first.
In fact, it's a strange trick of the AI boom that business leaders have spent this much time, money, and worry on AI without good data! We would never accept that blindness on lead gen, sales pipeline, or a key customer renewal. We ought to be able to say which AI work moved a business result we care about, with the data to prove it.
From there, decisions get easier:
- Sellers learn which skills to adopt, based on what works to sell more.
- RevOps sees whether a new vendor helps the team win.
- Enablement learns what to teach and who to teach it.
- The tools improve too, as the fruitful edits fold back into the shared version to lift the whole team.
That is the methodology, rooted in data, aimed at business results, and built to change how our teams get the job done. Map the usage, join it to outcomes, lift what works, and run it again.