When AI Sits In On Your Standup: The Fortune 500 Fight That Won’t End

(SeaPRwire) –

I caught up with Michael Chen, former Google People Operations senior director and long time workplace AI advisor, right after this week’s COO Summit, and he put this whole debate in perspective faster than any panel I sat through. This “colleague vs tool” fight isn’t semantics. It’s just the visible proxy for a broken management system most companies are still avoiding fixing. We’ve poured billions into building and buying AI tools for work, but barely spent a dime rethinking how we count work, budget for work, or reward people for working alongside AI. What you name your bot doesn’t change anything. Rewriting 100 year old management frameworks does.

This fight that just played out between two Fortune 500 veterans lays all these tensions out for everyone to see. Okta president and COO Eric Kelleher has already given proper names to every AI agent on his team, Leo, Sloan, Hank, Walker, and more. They show up to business reviews right alongside human team members. He said the turning point was a simple standup exercise where he asked everyone to name their own AI agents. Once people did that, AI stopped being just a tool and became part of the team, a shift he says made integration far easier. Just a few hours later, Cisco’s EVP and chief people officer Francine Katsoudas pushed back hard on that framing. She says AI belongs in your workflow, but it’s not a colleague, and clarifying that line early makes human workers far more confident about the shift.

What both sides, and every other leader at the summit, agree on is that every large company is stuck right now. Most have figured out how to run small AI experiments, but almost none have actually redesigned work around AI at scale. New data Cognizant presented at the summit puts hard numbers to that problem: 93% of all jobs are already disrupted by AI, six full years ahead of the 2023 projections the firm made. But the promised productivity gains everyone expected haven’t materialized. Researchers call that gap the “activation gap”.

Katsoudas opened up about Cisco’s own growing pains, when the company cut 4,000 jobs as part of an AI focused restructuring. She said teams that adopted AI the fastest actually saw internal trust drop about nine months in. Right now, the company is focused on investing in upskilling and internal redeployment instead of just handing out severance. In past restructurings, that approach helped 75% of impacted employees land new roles internally, and the team is working to push that number higher.

Multiple independent studies back up the risks of mismanaging AI integration. A randomized experiment published in Harvard Business Review this spring found that humanizing AI shifts accountability away from individual workers, leads to more unnecessary escalations, and reduces the quality of final human reviews. A separate BCG experiment found workers respond to AI colleagues by scapegoating the bot when things go wrong and getting more careless with their own work. University of Arizona research adds another layer: companies are stuck in a transparency trap. Disclosing AI use hurts colleague trust in the short term, but hiding your AI use and getting caught later is even worse.

Lattice CEO Sarah Franklin, whose company builds people management software, argues the solution is clear, blunt governance. She compares letting random AI tools into your company to letting a stranger into your home. You don’t let anyone in without asking who they are and why they’re there. The same rule applies to AI, you build clear guidelines and guardrails before you bring agents in, not after. Kelleher sees the problem from the other side. He says the real issue isn’t that human workers will feel displaced by named AI agents, it’s that managers still refuse to treat AI as a legitimate part of the workforce. For decades, managers have been trained to only care about headcount, org charts and reporting lines. That old way of thinking doesn’t work when AI is part of every team. His fix is to push AI token budgets down to frontline managers, forcing them to openly account for how AI fits into their team’s work and making that tradeoff visible in the budget. Franklin agrees with the core diagnosis: old performance management processes, the once or twice a year cyclical reviews that don’t keep up with shifting business priorities, are already broken. AI didn’t fix it, it just exposed how broken it is.

Under all the framing fighting, both Kelleher and Franklin agree on the core problem: the real bottleneck isn’t the AI, it’s management. Org charts, budget cycles, performance reviews, all of it was built for an all-human workforce, and none of it has been updated for the mixed workforce we have now.

What this fight tells us about where we’re going next is pretty clear. We went from talking about AI as a distant future to having AI agents in every team meeting in less than two years. Management systems haven’t even started to catch up. Cognizant’s analysis of 80,000 different work tasks found 90% still need some level of human involvement, so we’re not getting a fully AI workforce any time soon. We are going to have to rebuild almost every core management process over the next few years. We’ll see more teams test models that bake AI into formal budgets and team planning, just like Kelleher wants, and more companies put strict access guardrails in place like Franklin and Katsoudas call for. The end goal isn’t to pick a side in the colleague vs tool argument. It’s to accept that planning for a fixed number of human heads doesn’t work anymore, and shift to planning work itself, regardless of who or what is doing it. Right now, that’s a huge leap for most teams, but it’s a leap we have to make.

For this story, journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.