In this edition of Ashling Unfiltered, we put McKinsey’s Agents, robots, and us: Skill partnerships in the age of AI under the microscope. We pull the headline claim, check it against field reality, and end with one move you can make this quarter. Let's dive in.
In their latest research, McKinsey makes a critical observation about the current state of automation. They argue that true economic value doesn’t come from 1:1 task substitution, but from "skill partnerships" where humans and machines collaborate based on their comparative advantages. We agree. The "task-based approach" is exactly how to land in the "POC Quagmire."
More recently, organizations have been treating AI as a magic wand to wave over existing, broken processes. This might lead to isolated wins, but fail to scale. The future belongs to those who reimagine workflows entirely. As we move from a binary world (Human vs. Machine) to a multi-actor ecosystem, agentic AI orchestration becomes pivotal. A redesigned workflow today must account for this complex cast of characters:
Redesigning an agentic automation workflow means asking: If we have these four actors available, would we even build the process this way in the first place? Sometimes, the answer is no. You don't need to automate the old way; you can architect a new way.
At Ashling, we design for this new multi-actor reality. AI agents change the rules of process design and agentic delivery, which means the goal is no longer to automate legacy workflows but to architect better ones. By orchestrating humans, robots, and agents from the start, we help clients re-engineer workflows for the agentic future that unlock new levels of speed, quality, and control.
The Task-Based Approach
A traditional automation strategy focuses on isolated steps. For example, a Customer Care agent might still need to manually save claim submission attachments via email to trigger downstream processes. While traditional automation can extract data and enter into the CMS, the workflow hits a wall: a queue item is created for a human adjudicator. The automation saves keystrokes, but the adjudicator is still stuck manually researching and verifying the provider. The data entry task is faster, but the workflow remains disjointed.
The Reimagined Workflow
A redesigned agentic automation strategy focuses on a continuous, intelligent stream. An intake AI agent monitors email traffic, automatically opening, classifying, and routing all documents from a shared inbox for processing—removing the Customer Care agent from the loop entirely. Depending on the attachment type (prior authorization, claim, predetermination, eligibility certification, explanation of benefits, etc.), an AI agent can trigger the corresponding downstream process, calling on specialized AI agents to carry through the request.
For example, the Provider Verification Agent instantly validates clinician practices, licenses, and misconduct by searching across the web, internal and external databases. It delivers the results to the adjudicator, equipping them to make a more informed decision when reviewing the claim.
Move into the age of Skill Partnerships by doing this now:
- Treat AI as a Hammer, Not a Universal Tool: Job items/tasks in a process are like nails and screws. Some require a hammer, and others, a Phillips screwdriver. In other words, AI is one tool in the toolbox, not the whole kit. Instead of asking how to layer AI onto your current workflow, ask whether AI is the right fit for the job. If the process remains identical after AI adoption, you likely haven't maximized its value.
- Audit for "Islands of Automation": Look for places where you have successfully deployed a bot or AI tool, but the upstream and downstream steps are still manual bottlenecks. These are prime candidates for end-to-end workflow optimizations.
- Build with Organizational Design in Mind: Agentic workflows are not just a technology change, they are an organizational design change. When robots, agents, and humans work together, you may need to adjust roles, skills, and ways of working to support these new capabilities.
- Look to the Past: Revisit automations that were previously deemed technically non-feasible. With agentic capabilities in the mix, some of those candidates may now be viable.
If you’re looking to learn more about AI and Agentic processes that have been deployed to production and the realities of what it takes to get them there—bring Ashling in for an education and knowledge sharing session. We offer hands-on workshops to help build your team’s understanding and uncover valuable, prioritized opportunities to kick-start your agentic AI journey.
.jpg?width=352&name=Open%20AIs%20Agent%20Builder_%20Do%20We%20Really%20Need%20Another%20Workflow%20Builder_%20(1).jpg)
