For the better part of the last century, the answer to more work has been more people. If a broadcast operation needed to handle more content or an ad tech team needed to traffic more campaigns, it hired. Technology is now a genuine alternative, and for the first time in decades, leaders of knowledge workers have a real choice about how work gets done. Our team sits on both sides of that choice, placing technical talent with companies and building AI agents that take on work people used to do by hand. From that vantage point, it’s clear that the companies getting the most from AI are the ones that have figured out how people and technology should work together, and then redesigned the work around that answer.
The work without a clear owner
Most companies approached AI as a technology rollout: pick a vendor, switch it on, then manage the change afterward. But that misses where the value actually comes from. AI pays off when a company intentionally changes its operating model, its processes, and the shape of individual roles around AI. That work doesn’t sit neatly inside IT, HR, or the business. Each has a piece of it, so no one owns all of it. Who owns AI transformation is one of the most important open questions in enterprise AI right now, and it plays out in nearly every client conversation.
Companies have also lost sight of work at the task level. Job descriptions and process maps describe roles too broadly to tell you which tasks an agent should take on and which ones need a person. When scoping an agent for a media or telecom team, the first step is always understanding what people actually do hour to hour. That’s where both the opportunity and the risk become visible.
Every automation creates an exception
Anyone who’s designed a manufacturing process will tell you not to discount the human element. Factories are the perfect example; they’ve been automating for decades, and yet they never eliminated the need for people to catch what the machines miss. Every automation creates an exception that needs supervision and human handling, and knowledge work is no different.
This matters a great deal in fields like broadcast and ad tech, where the work is high-volume, time-sensitive, and full of edge cases. An agent can clean a dataset, pull audience insights, or handle routine ad server tasks far faster than a person can, but a campaign with a conflicting brief or a data feed that suddenly changes format still needs someone with judgment. The goal is clear visibility into which tasks AI can absorb, without mapping every step so tightly that there’s no room left for the improvisation that makes good teams effective.
Sequencing matters, too. AI often gets dropped into an existing workflow, alternating with human steps in a way that adds friction instead of removing it. A well-designed process lets AI run several steps back to back, then hands off to a person at the points where judgment counts.
Automation moves work rather than removing it
Automating part of a role is the easier problem. What happens to the rest of the role is harder. People fill freed-up time quickly, and their jobs keep evolving as they take on new responsibilities. That’s why so many automation programs struggle to show a simple before-and-after impact on headcount.
For talent, this changes what companies should be hiring for. Technical roles increasingly require people who can work alongside agents, understand their limits, and step in when an exception surfaces. Critical thinking, domain expertise, and enough AI fluency to apply the tools well are becoming as important as any specific technical skill. That applies to managers as much as to individual contributors. If AI absorbs the entry-level tasks people once learned on, and managers aren’t fluent enough in the technology to guide what replaces that learning, companies will struggle to develop their next generation of talent.
Performance needs a new definition as well. When one person using AI well can produce what a small team once did, output alone stops telling the whole story. The judgment and accountability behind the result matter just as much.
People-first companies will get further
The bottom line for organizations is that technology and workforce decisions can no longer run on separate tracks. HR and IT leaders need to own this jointly, with the business defining which problems are worth solving. Companies that treat AI purely as an IT project end up with tools that don’t fit how their people work, while companies that treat it purely as an HR initiative lack the technical grounding to make good decisions.
Companies that are people-first, not tech-first, are the ones most primed to get workers collaborating with machines. Yet most organizations still put far more money into the technology than into the people expected to use it. That ratio needs to flip. The true design is people plus technology working together, and the companies that invest accordingly will be the ones that see a real return.