SCOTTSDALE, Ariz.—Executives gathered here at the Fortune COO Summit on June 1 didn’t just debate technology. They wrestled with a deeper question. Should companies treat AI agents as colleagues?
The room split. Some saw agents as the next members of the team. Others warned that label invites trouble. No easy consensus emerged. Yet one point rang clear. The way leaders frame these systems will shape everything from org charts to trust levels inside companies.
One COO’s Radical Experiment
Eric Kelleher, president and COO of Okta, has already crossed a line many peers still eye warily. He named his agents. Leo. Sloan. Hank. Walker. They appear in business reviews right next to human staff. “In that exercise, AI became a colleague as opposed to a tool and that catalyst is valuable,” Kelleher told the summit, according to a Fortune article.
He went further. Kelleher booked a flight to Bangalore. The entire trip he stood up an open-source agent on a separate machine. Then he assigned the same exercise to his leadership team. “That flight to me was transformative in how I recognized what the capabilities of this technology are,” he said in the same report.
His message lands blunt. Managers have spent decades fixated on headcount. Payroll. Who reports to whom. That model breaks when digital workers join the mix. Kelleher pushes for explicit budgets that cover both human labor and digital labor. He calls it a shift from workforce planning to work planning. “We have trained every manager in the world to think about one thing, and that is: what’s their headcount?” he explained. The leap feels big. Uncomfortable. But necessary.
And he isn’t alone in seeing momentum. Recent discussions on X show practitioners already building teams of named agents that handle real workflows. One post from June 2 highlighted Salesforce assigning official responsibilities to agents and tracking their performance like any other contributor.
Yet the counterarguments hit hard. Francine Katsoudas, EVP and chief people, policy and purpose officer at Cisco, drew a firm boundary. “I would not look at AI as a colleague. I think we should look at AI and agents as part of the workflow, but not a colleague. And I think the sooner we land that, the more confident our people will be,” she stated at the summit per the Fortune coverage.
Cisco’s experience adds weight. The company restructured with AI and cut 4,000 jobs. Trust dropped in teams using the new systems after about nine months. Katsoudas stressed heavy investment in skills and redeployment. Past efforts placed 75% of affected workers. The goal now sits higher. Clear guardrails matter. Treat agents like visitors with strict rules. Anything less risks confusion.
Research backs her caution. A May 2026 Harvard Business Review study found that humanizing AI shifts accountability away from people. It increases escalation and reduces the quality of human review. Workers scapegoat the agent. They grow careless. Boston Consulting Group experiments showed similar patterns. University of Arizona findings revealed that disclosing AI use erodes trust in the short term. Staying silent and getting caught proves worse.
So the semantics carry consequences. Call it a colleague and managers may over-delegate. Call it a tool and teams might underuse its potential. The summit exposed this tension without resolving it.
Productivity gaps complicate matters further. Cognizant shared data at the event. AI will disrupt 93% of jobs by 2026, faster than 2023 forecasts predicted. Yet an activation gap persists. Analysis of 80,000 tasks showed humans still needed in 90% of them. Productivity gains haven’t materialized at scale. Sushant Warikoo of Cognizant put it plainly. Humans and agents hold equal privilege in some setups. The point remains amplifying human potential.
Jon Blotner, president at Wayfair, took a different tack. He reversed a top-down AI mandate. Teams gained open access to Claude, Gemini and ChatGPT. Roles began to reinvent themselves. “We see people reinvent their jobs,” he noted. That person stays incredibly valuable.
Sarah Franklin, CEO of Lattice, pointed to broken performance management. Reviews happen once or twice a year. They feel disconnected from actual business movement. AI exposes the gap between OKRs and daily priorities. The systems don’t match reality anymore.
These disagreements reflect larger forces. Companies race to adopt agents. Many still deny the full redesign required. Kelleher compares the moment to electricity’s arrival. Factories clung to steam engines too long. Today’s chatbot experiments risk the same slow realization.
Recent coverage adds texture. A UNSW BusinessThink article from April 2026 examined Atlassian’s Rovo agents. Marketed as configurable AI teammates inside Jira and Confluence, they automated millions of workflows. Yet the company later cut 10% of its workforce to fund more AI investment. The juxtaposition raises hard questions about coexistence versus replacement.
Microsoft’s Work Trend Index from May 2026 highlighted frontier professionals who already treat agents differently. They brainstorm processes together. They document handoffs and quality standards at team, function and organization levels. These habits separate high performers from the rest.
Even hardware signals the shift. NVIDIA’s recent push for AI-focused PCs aims to run advanced agents locally. The company positions the laptop itself as a teammate rather than a device. Microsoft redesigns Windows around agents. Creative tools follow.
But practical risks remain. Over-reliance without governance creates brittle systems. Under-reliance leaves money on the table. The COO role sits at the center. These leaders must translate experiments into policy. They must balance speed with stability.
Kelleher’s team now sees digital agents as colleagues of sorts. The framing changed behavior. Yet Katsoudas insists on workflow language to protect people. Both views carry truth. Neither fully satisfies.
Executives left Scottsdale without a single answer. They carried sharper questions instead. How do you budget for invisible labor? What accountability rules apply when an agent errs? How do you maintain trust when humans and code share the same standup?
The debate won’t fade. Agents grow more capable by the month. Companies that settle the colleague question too quickly may regret it. Those that ignore it risk falling behind. The real work starts now. Redesigning tasks. Rewiring management habits. Accepting that the org chart of 2026 looks nothing like the one from 2023. And doing so without losing the humans who still drive judgment, creativity and accountability.
WebProNews is an iEntry Publication