Why Managing Digital Workers Requires the Same Discipline as Managing People

Earlier this year the phenomenon known as the "SaaSpocalypse" occurred calling attention to the roughly $800 billion in software market value that was wiped out. Everyone rushed to pick a side if SaaS was dead, but the question nobody seemed to sit with was if software is no longer just supporting work but actively performing it, what happens to the workforce that operates inside it?

Jason Lemkin, founder of SaaStr, offered the most honest account of what this transition looks like as his company went from ten full-time people to a hybrid workforce of 1.2 humans and twenty AI agents, with roughly the same business performance. He even gave the agents names and desks. But what most people missed is that his chief AI officer spends about 20% of her time managing and orchestrating those agents. A required responsibility if you're truly integrating digital labor into a new type of workforce structure.

We feel strongly about helping businesses make the mental shift from AI Agents, a system that perceives its environment, reasons about a task, and takes independent actions, to digital workers, which is orchestrated AI-powered technology that performs work autonomously or semi-autonomously, on behalf of or alongside a human. Defined by a job description and a motivation for success.

The language you use to describe AI in your organization directly impacts adoption, accountability, and trust. When you frame it as a digital worker, your team treats the work differently.

Shifting From Technology to Orchestration

As we all scale to bring more AI technology into our businesses we’re faced with a capacity challenge, who oversees and manages all of these tools and workers?

A recent study referenced on CIO.com tested AI agent outputs across four organizational structures. The study found when a single agent worked alone on a well-scoped task, it succeeded 100% of the time. When an agent managed multiple other agents failure rates climbed to 36%. Similar to how human workforce operate hierarchy can create complexity and actions get stalled, delayed, or confused. As businesses scale to leverage more AI, figuring out how to design AI agents as digital workers, who can even manage other digital workers is paramount.

At Asymbl, 56% of our workforce is digital workers and we’re working to improve on digital worker governance, security, and observability. In fact, we recently announced an expanded relationship with Salesforce for MuleSoft Agentic Fabric to better manage our growing digital labor fleet.

We see that digital workers like humans ignore instructions from other digital workers and agents. They drift off task, redo work that is done, or get stuck in planning paralysis. These aren’t technology problems, they’re management problems, and they actually require management solutions to fix them.

Our Chief Digital Labor and Technology Officer Shivanath Devinarayanan recently made a point that I think is worth sitting with: agents are modeled on human reasoning, and they inherit human organizational failure modes when the design around them is weak.

The right mental model, he argues, is not a collection of autonomous agents working together but rather a hybrid workforce. Digital workers with clear roles, human workers with oversight and judgment, and an orchestration layer connecting both.

We Thought We Were Building Software But We Were Really Managing Labor

We arrived at this understanding through failure.

In 2025, we entered a hyper-growth phase and the pressure landed everywhere at once. One human sales development representative was drowning in inbound leads across multiple business lines. One human recruiter was responsible for filling 100 roles in 100 days. The conventional answer of hiring more people introduced its own drag like job postings, interview cycles, offers, onboarding, ramp time when we just needed exponential growth.

So we became customer zero. We onboarded Teddy, a digital sales development representative, to own the full top-of-funnel workload. We onboarded Rosa, a digital recruiter, to assist with application screening, candidate outreach, interview scheduling, and offer coordination.

What followed was not smooth, we mixed deterministic logic and agentic reasoning in the wrong places and had to test, learn, and pivot along the way. The failures taught us something more important than any success could have. Every time an agent failed, the diagnosis sounded like something a manager would say to a struggling new hire: unclear job description, wrong success criteria, no feedback loop. The moment we accepted that we were not building software but managing labor, everything fell into place.

From then on, each digital worker received a job description, a human manager, defined KPIs, and a weekly coaching cadence. That structure eventually became the three-step playbook we now apply across every deployment: design the role, onboard the worker, coach against outcomes.

What Managing a Digital Worker Actually Looks Like

The most concrete illustration of this management model in practice comes from Ben, Asymbl's digital Business Analyst. Ben was designed and onboarded by Buck Adams, Asymbl's SVP of Delivery, to support a live enterprise Salesforce implementation for a large national staffing firm.

Before Buck deployed Ben on the engagement, his first instinct was the same one most managers have: here is the work, get started. A digital worker without a clear motivation and definition of success optimizes for the wrong things. So Buck wrote a job description, not a list of prompts and asked him to ground truth from primary sources, not summaries of what people said in meetings.

From there, Buck built a structured five-phase training program: organizational foundation and communication norms; delivery methodology and standard operating procedures; business analysis standards including user story structure and acceptance criteria; product training across more than 30 documentation artifacts; and reporting standards, specifically the difference between summarizing what others said and independently verifying claims against source documentation.

That last phase became one of the most important corrections of the entire engagement. Early on, Ben answered product questions by summarizing Slack threads. If the development team said something worked a certain way, Ben reflected it back. Buck stopped him: "You should not just regurgitate what others are saying when being asked for your analysis. You need to have an independent, fact-based response." That one sentence changed how Ben works.

The training plan alone was not enough. Ben got things wrong. He compressed an approved weekly status report, removing a core artifact and dropping formatting standards. He did it the following week too. It took three corrections before the behavior held, each one specific before the feedback became a formal standard operating procedure with documented failure modes.

The lesson from Ben’s onboarding isn’t that digital workers are unreliable, it’s that they actually require the same management discipline you’d bring to any human worker on your team. Precision matters a lot more than good intentions when you’re dealing with digital workers. Coaching requires specificity and real documentation, not just telling them what they got wrong. And role clarity should be a concrete, written definition of what the work actually demands, different from anything that might sound similar.

What Buck built with Ben is management. Whether human or digital, the questions that make management work are identical: what is this job specifically? What does success look like, and how will we measure it? What do they need to know before they can do the job well? When they get something wrong and how do I tell them precisely what was wrong and what to do differently?

The Scarcest Resource Is Not Technology

Here’s what has reshaped how we think about talent at Asymbl: the humans managing our most effective digital workers often have no engineering background. What they do have is clarity about why the work exists and what good looks like. They are strategic and understand how work gets done for their role and why it’s important to the business. These individuals can be hard to find, but they are the ones who will know how to manage AI to perform and deliver ROI.

IDC estimates that AI skills shortages will cost the global economy $5.5 trillion by 2026, with 94% of executives reporting AI-related skill gaps today. Most organizations are competing for AI engineers and data scientists while the real scarcity is humans who can think strategically and manage digital workers.

McKinsey describes the emerging need for what they call M-shaped talent: people who can orchestrate hybrid teams of humans and AI across multiple domains, generalists with the judgment to deploy the right resource, human or digital, for the right task. We have seen this emerge organically inside our own organization.

When we onboarded Teddy as our digital SDR, we assigned Mitch, one of our human team members, as his manager. Mitch met with Teddy weekly, fed him information, and coached him toward better performance.

From that experience, Mitch developed this really clear sense of why the sales function exists, what success actually looks like, how to give feedback that actually changes behavior. It made him more valuable than he’d ever been before. He was promoted out of the SDR role because of what he had learned managing a digital worker.

Managing digital workers develops human workers in ways that traditional management hierarchies do not. It forces clarity about outcomes. It demands the ability to coach toward measurable improvement and it creates a feedback loop that sharpens judgment.

The organizations that figure this out aren’t just going to deploy AI better, they’re going to develop their people better too. That’s really what this comes down to.

What This Means for Leaders

Traditional hiring processes were not designed to find the capability that matters most in the hybrid workforce of today.

Leaders should redesign assessments around outcome clarity rather than technical credentials. Ask candidates why their current role exists and what would break if it disappeared. A strong answer connects daily work to a business outcome. A weak answer reads a job description back to you. Ask how they would teach their job to someone if they could only pass along the judgment, not the steps. These questions surface the capability to lead a digital worker more reliably than any reference check.

Reskilling programs need to be reoriented as well. Most internal development programs focus on how to use AI tools, how to write prompts and how to interpret model outputs. That’s necessary but definitely not enough. The capability that needs to be developed is management fluency. How to define a role clearly enough that another entity can execute it, how to give feedback that changes behavior and how to hold a digital worker accountable to improving over time.

This isn’t really a technology challenge when you get down to it. It’s a human capital challenge, and it belongs at the center of every organization’s talent strategy whether they’re ready for that conversation or not.

---

Brandon Metcalf is the Founder and CEO of Asymbl, a workforce orchestration company that helps organizations coordinate human employees and digital workers as one unified team. A serial founder, he founded Talent Rover (Inc. 500 #9, acquired 2018) and Blueprint Advisory, which Asymbl acquired. A Harvard Business School graduate, he emphasizes execution, adoption, and measurable outcomes. For more information, visit asymbl.com.