Agentic AI Has a People Problem

We’re deep in positioning work for a client building an agentic AI platform. I won’t name them (yet), but the product is impressive. Agents that catch freight exceptions before they compound, renegotiate contracts in real time, and book appointments without anyone asking.

There’s no playbook for positioning something like this. You can’t pull the standard SaaS go-to-market off the shelf and apply it to software that makes decisions on its own. The messaging, the visuals, the brand story… nobody’s written it yet. Because the category is too new.

What most companies do instead (including a lot of smart ones) is lead with what the product can do. Big capability claims. Efficiency numbers. Autonomy as a selling point. That works for the executive writing the check.

It terrifies everyone else.

Agentic AI doesn’t assist. It acts. That changes everything.

When spreadsheet software arrived in the 1980s, people in finance braced for the worst. Some clerical jobs did disappear. But most people who worked with numbers found that Excel multiplied what they could do. The anxiety was real. So was the upside. That’s why the story resolved.

Every enterprise tool before this one waited for a human to operate it. You ran the query. You read the output. You made the call. The software was powerful, but passive. It needed you.

Agentic AI doesn’t wait. Vendor risks get resolved without being handed off. Shipping exceptions get handled before anyone flags them. Appointments get rebooked automatically. The human used to be the last step in the sequence. Now the software is.

That changes how people experience the product, especially the ones whose work is changing most. 51% of workers are already worried about losing their jobs to AI this year. Among frontline workers, only 18% feel their jobs are secure.

Most companies selling agentic AI are leading with messaging that reads like a threat.

Your product has three audiences. Most companies write for one.

The executive who bought the product is sold. They saw the demo, ran the numbers, signed the check. They want efficiency, cost reduction, and competitive positioning. The ROI story lands because they’re not the ones whose work is changing.

The department head implementing it has different concerns. Change management. Integration. Whether their team will actually engage with the product or quietly route around it. They need to know how to bring their people along, not just what the product can do.

Then there’s the user. The person who found out about this in an all-hands and is now trying to figure out what it means for their job. They’re not asking about ROI. They’re asking if they’re going to be okay.

Most companies write one message. The executive message. “A limitless, skilled digital workforce.” “Take action without human intervention.” “Autonomously resolve 80% of common customer service issues without human interventions.” These claims might be accurate. But from the analyst’s desk, it reads like a layoff notice.

Workers anxious about AI are 45% more likely to disengage. A disengaged user doesn’t revolt. Worse, they comply minimally. Bad data goes in. Edge cases are ignored. Errors are uncaught. The ROI the executive paid for disappeared, and nobody can figure out why the technology isn’t delivering.

Your biggest potential champions are the ones you’re alienating

The second mistake companies make when they realize they have a messaging problem: they add an augmentation line.

Every major agentic AI company says some version of it. Salesforce says the digital labor model is “not about replacing the human workforce but augmenting capabilities and freeing up time for more creative and strategic work.” ServiceNow promises to “free employees from the mundane, repetitive, and tedious tasks.” UiPath says agents let humans “focus on higher-value work.” But when 63% of workers say AI will make the workplace feel less human, that message is landing everywhere except with the people it’s meant for.

What does higher-value work mean for this person, in this role, on this team? What specifically changes about their day? What do they get to spend time on that they couldn’t before? Nobody is answering those questions.

The framing that works with all three audiences

Back to our client. We went beyond talking about what the agents could do and started talking about what they meant for the people already doing the work.

Your team has enough complexity to manage. They shouldn’t have to manage all of it manually.

Put the human at the center of the value, not the machine. The agent handles volume. The human handles judgement. The agent handles routine. The human handles the unprecedented. The result belongs to the team, not the software.

That structure lets you speak to all three audiences without contradicting yourself. The executive hears efficiency. The operator gets a change management story they can actually tell their team. The user hears something they almost never hear from an AI company: their judgement is what makes this work.

It targets the work, not the worker. “Nobody should spend their week chasing carrier updates manually” is an indictment of a process, not the specialist running it.

We’re beginning to build a repeatable structure. The harder question is whether companies building agentic AI will use it before the window closes.

It gets harder as the tech gets smarter

The agents shipping today are scoped and well-defined. The ones arriving in 18 months will be broader, more autonomous, and harder to explain to the person whose work they’re changing.

Companies that build a coherent human-centered narrative now will have it when they need it most. The ones that keep leading with capability claims will find themselves retrofitting trust onto a product people have already decided to distrust.

When every agentic AI product converges on similar benchmarks (and they will) trust will become the only thing left to compete on.

Trust is a brand problem. And that’s great news, because brand problems have real solutions. The companies that recognize this first will be very hard to displace.

This is the work we do at Matic. If you’re building or selling agentic AI and your positioning is overlooking the end user, let’s talk.