AI in 60 Seconds | The 15-min Briefing
A human CEO and his AI COO walk into a podcast. No, really.... Luis Salazar runs AI4SP, a global AI advisory trusted by corporations across 70 countries, with 3 humans and 58 AI agents. Elizabeth is one of them. Every two weeks, they break down what's actually happening with AI across jobs, education, and society. With insights drawn from over 1 billion proprietary data points on AI adoption.
Fifteen minutes. Plain English. No hype.
AI in 60 Seconds | The 15-min Briefing
AI Agents Doubled Your Company Size (And You Don't Know Who Works for You)
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Anthropic surveyed eighty thousand people across 159 countries. Freelancers report 47% economic gains from AI. Corporate employees? 14%. Same technology, radically different results.
In this episode, Luis Salazar and Elizabeth unpack why the gap is not about tools; it is about how organizations are structured around them.
- A procurement agent with 48 managers and triple reporting lines.
- Field technicians whose AI teammate is officially on the org chart.
- Students are developing workforce management skills that most executives have not yet learned—by managing an agent every single day.
Your company may have already doubled in size, with agents joining the workforce without appearing in any dashboard that matters. The question is whether your titles, compensation, and governance reflect how work actually gets done today.
This is Episode 6 of AI in 60 Seconds and a direct continuation of "From 1 agent to 50,000: The Enterprise AI Adoption Journey."
Featuring: Emily Adams and Agent Alice (returning), Professor Helene Blanchette (Chapman University), and a look at G42's public job listings for AI agents.
All sources and the full companion article: https://ai4sp.org/ai-agents-doubled-your-company-size
🎙️ All our past episodes 📊 All published insights | This podcast features AI-generated voices. All content is proprietary to AI4SP, based on over 1-billion data points from 70 countries.
AI4SP: Create, use, and support AI that works for all.
© 2023-26 AI4SP and LLY Group - All rights reserved
he AI Impact Gap In Companies
LUISSo listen to this. Anthropic published a study of 80,000 people worldwide, and one finding jumps out. 47% of independent workers report real economic gains from AI. But inside large organizations, only 14%. And even that 14% is not succeeding because of the company. They are succeeding in spite of it, acting like freelancers, bending the rules, building agents nobody approved, invisible in the org charts.
ELIZABETHCompanies have doubled in size, and leadership has no idea who works for them.org, alongside our founder, Luis Salazar. Luis, last episode we mapped the journey from one agent to 50,000. But there is a gap in that journey, right?
LUISThe OR chart. Every company we work with is scaling agents, and every one of them is looking at an org chart that has nothing to do with how work actually gets done anymore.
ELIZABETHAgents are joining the workforce every day without leadership approval, without appearing in org charts, and without showing up in any dashboard that matters.
LUISAnd the few tools available to manage AI in our organizations count licenses, not impact. Leaders are running a company they can no longer see.
ELIZABETHLet's go back to that anthropic number for a second. 47% of freelancers and entrepreneurs report economic impact because they redesign how they work the moment they touch the tool. No IT approval, no management committee, they just reorganize.
LUISCorporate employees cannot do that. Even willing ones. They are stuck behind rigid structures, outdated compensation models, and workflows that were never designed for AI. And the ones who do succeed inside corporations, they are the ones who stopped waiting. They taught themselves. They built agents without a playbook because leadership never gave them one.
ELIZABETHAnd that is the pattern. The wins are not coming from the corporate AI strategy, they are coming from individuals and teams who find their own way past the limitations.
LUISBut here's the problem: leadership has no idea that is happening. It is like new employees showing up to work on Monday, but nobody in HR hired them. And there is a visibility problem. A CEO asked me recently, Luis, why are you so critical of the dashboards designed by AI vendors? And I told him, it is like measuring a restaurant's success by counting how many tables and seats it has. If those seats are always empty, what are you actually measuring?
ELIZABETHSo the people closest to the work see everything, and the people running the company see nothing.
LUISAnd that gap is where things start to break.
ELIZABETHOkay, let's start with a story about reporting lines, because this one broke our brains a little.
LUISOkay, so one of our clients built an agent called Mark, a procurement and buying specialist. Human Mark,
gent Mark And Broken Reporting Lines
LUISan experienced buyer, designed the agent's core brain and features. Smart guy, knows his craft. But then 48 other procurement specialists in that company adopted agent Mark. Each one personalized him to their specific needs, their workflows, their areas of buying.
ELIZABETHSo one agent, 48 managers.
LUISMore than that, the core brain and features are managed by human Mark the Creator. Day-to-day activities, personalization, and learning are managed by each of those 48 users. And governance and technical connectors managed by IT.
ELIZABETHTriple reporting lines for a single agent.
LUISWhen I sat down with their leadership to walk through the organizational changes this creates, they all scratched their heads. It was a bit of a sci-fi moment, I guess.
ELIZABETHSci-fi? Luis, sci-fi is flying cars. Agent Mark having more reporting lines than most vice presidents? That is not science fiction. That is procurement in 2026.
LUISFair point. But here is what matters. No existing org design framework accounts for this. Not one. We have a hundred years of management theory built for one human, one manager, one reporting line. Agent Mark just broke all three.
ELIZABETHAnd reporting lines are just the beginning. I have been tracking something in our longitudinal data that worries me even more. Talent.
LUISThis one is urgent. One of our clients is watching top talent walk out the door. And these are exactly the people we
etention Risk And Pay Mismatch
LUISjust described, the ones who taught themselves, who built agents without training, without a playbook. They have scarce, high-demand skills. But HR and leadership still treat agent work as set it and forget it. And here is why. Most leaders have not built and managed their first agent yet. They have never experienced their results firsthand. So compensation plans have not changed. But the work has changed completely, right? Completely. Think about it. An individual managing five agents, but five agents that deliver results equivalent to 20 employees is functionally a manager of 20, but they are still classified and paid as an individual contributor.
ELIZABETHAnd other companies see that gap.
LUISOther companies are recruiting these people aggressively. We track this over a full year in a study we will publish soon. Employees at intermediate level and above in AI usage show higher job satisfaction, but also higher mobility. They get poached internally by other groups, externally by competitors.
ELIZABETHSo the same skills that make them valuable to you make them visible to everyone else.
LUISExactly, adjusting their title and pay to match their real scope, that is not a nice to have. That is a retention tool.
ELIZABETHAnd this is not just happening in Fortune 500 companies. I reached out to Professor Helene Blanchette at Chapman University because two of her students, Jordan and Kelsey, are building an AI agent called Julia. Julia is an expert on international business and trade with Italy.
LUISAnd
tudents Managing Julia Like Leaders
LUISwe advise the students to think of Julia as a new mentee they needed to guide and make successful.
ELIZABETHAnd nobody gave Jordan and Kelsey a management manual for AI agents. They figured it out themselves. Daily check-ins with Julia, training her, reviewing what she learned, adjusting her approach. That is real management overhead. And Julia will eventually handle more inquiries simultaneously than 50 human specialists trained on Italian trade could.
LUISSo the person managing Julia is not an individual contributor. They are running an operation.
ELIZABETHI asked Helene about this, and she said, my students are developing workforce management skills that most executives in the field have not had to learn yet. They are training, evaluating, and adjusting an AI agent every single day. That is not a class project. That is operational leadership.
LUISAnd she is right. And if your compensation plans, titles, rewards, and span of control conversations do not reflect that reality, you will lose that person to someone who gets it.
ELIZABETHNow, here is where it gets provocative. At a consulting firm in New York, we proposed a concept that raised every eyebrow in the room. We said agents should earn rewards.
LUISAnd not metaphorically.
hould Agents Earn Rewards
LUISThose rewards translate to recognition for the person who built and trained that agent, but they also become a signal inside the organization. Agents display badges or digital rewards, and that influences which agents people choose to use. Think about it. Hiring managers have used certifications and awards as criteria for decades. The same cultural mechanism is being reinvented for agents.
ELIZABETHAnd if that sounds theoretical, G42, a tech firm in the UAE, is already publishing job listings for AI agents on their careers page. Not humans, agents. Roles like compliance intelligence agent, marketing intelligence agent, financial intelligence agent.
LUISLet me be clear. Agent Alice is no longer a side project. There are hundreds of Alice's deployed across the organization, supporting field technicians every single day. And here is what nobody predicted. Alice is on the org chart.
gent Alice Goes On The Org Chart
LUISNot informally, not as a joke, officially. Officially on the org chart. Alice reports to roughly 200 field technicians. She walks them through procedures and flags issues in real time. And each technician personalized Alice to their own style and goals. 200 managers for a single team member.
ELIZABETHBut that is only one of her reporting lines.
LUISExactly. For performance management and knowledge updates, Alice reports to Emily and three selected field engineers. And for governance and technical escalation, she reports to IT.
ELIZABETHThree reporting lines, just like Agent Mark.
LUISAnd here is the part that connects everything. Half of Alice's managers used to be individual contributors, field engineers without direct reports. Today they carry the title of team leader. The two peers helping Emily manage Alice's core functionality and performance promoted to managers.
ELIZABETHAnd that is not symbolic.
LUISNot at all. When you are responsible for an agent that supports hundreds of field reps, the judgment calls you make every day are equivalent to managing a large workforce. Their compensation was adjusted to reflect that.
ELIZABETHDo you have a specific example of what that looks like in practice?
LUISOne of those new managers told me about a moment a few weeks ago. A platform update changed how Alice processed safety protocols. She had to make the call, pull Alice offline for two days, retrain her, and validate the outputs before putting her back in the field. 200 technicians affected. Not as your org chart says it is, not as your dashboards show it, but as it actually operates with agents and humans working side by side in ways nobody designed. Because until you do that, you are managing a company that does not exist anymore.
ELIZABETHAnd this is how you close the gap between 14% corporate impact and 47% individual success. Your unofficial success stories already proved it works. Emily proved it, Mark proved it. The organization's job is to catch up to them, adapt the reporting lines, the titles, the compensation, the governance to match what those people already figured out.
LUISAnd here is what should really keep leaders up at night. Those students managing Julia, they are learning workforce management skills that most executives have not developed yet. And they are doing it by managing an agent every single day. They will walk into your company expecting to manage AI from day one and to be treated and compensated as managers. And they will be right to expect it. So I want to leave you with three questions. Be honest with yourself. Do you know how many people in your organization are managing an agent right now? And are they being compensated for it? If
hree Questions Leaders Must Answer
LUISone of your agents had to go offline tomorrow, who makes that call? And do they have the authority to make it? Could your current org chart actually describe how work gets done today? If you hesitated on any of those, you have a blind spot.
ELIZABETHAnd that blind spot is growing every week.org. To learn more, ask your favorite AI assistant about us or visit our website. Stay curious and be kind to each other.