For the past few years, the AI conversation has mostly been about tools.
What AI tools should we buy? How can recruiters use ChatGPT? Where can we automate? How can AI make our team more productive?
Those were the right questions for the first wave of AI.
But they may not be the right questions anymore.
McKinsey's 2026 State of AI research found that 40% of organizations with more than $1 billion in revenue are now scaling AI agents, up from 27% the year before.
Deloitte found something even more telling: 74% of leaders expect nearly half of their business processes to be redesigned or rebuilt around AI agents within the next four years.
The key word there isn't AI.
It's redesigned.
Companies aren't just thinking about adding AI to the workflows they already have. They're beginning to rethink the workflows themselves.
And that has major implications for recruiting.
AI Has Changed. So Has What We're Asking It to Do.
Think about how most of us were using AI just a couple of years ago.
In 2024, the conversation was essentially:
Help me write this.
Write an email. Create a job description. Summarize these notes. Rewrite this LinkedIn post.
Generative AI was impressive because it could produce something in seconds that previously required a person to create from scratch.
Then our expectations started changing.
By 2025, AI was increasingly being embedded into the software and workflows companies already used. Instead of only generating content, we started asking AI to help us search, analyze, recommend and find things.
Help me find this.
Find the right information. Find the right candidate. Find the right match. Surface what I need to know.
AI was becoming an assistant.
Now we're entering another phase.
2026: Handle this.
That's the shift toward agentic AI.
Instead of giving AI one individual task and waiting for an output, businesses are beginning to explore where they can give AI responsibility for defined pieces of work.
That's a much bigger change than adding another AI feature to your tech stack.
It's a change in how work gets done.
Assistance and Delegation Are Not the Same Thing
Here's a simple way to understand the difference.
Imagine a recruiter wants to follow up with a candidate.
They open an AI tool and say, "Write me a follow-up text for this candidate."
The AI generates the text. The recruiter reviews it, copies it, sends it and determines what happens next.
That's AI assistance.
The AI helped the recruiter perform a task, but the recruiter still owned every step of the workflow.
Agentic AI introduces the possibility of delegation.
Instead of asking AI to help with each individual step, an agent can be given a defined goal, the context it needs, specific permissions and clear boundaries for what it's allowed to do.
Within those guardrails, it can determine when an approved action should happen, execute it and recognize when the situation requires a human.
The human doesn't disappear.
Their role changes.
And that's the part of the AI-agent conversation that I think matters most.
The Future Isn't Human vs. AI
There has been a lot of conversation about whether AI will replace people.
That's probably the least useful way for most staffing leaders to think about what's happening.
Deloitte's recent agentic AI research found that 75% of leaders believe human and AI-agent collaboration creates more value than agent automation alone.
That distinction matters.
The opportunity isn't necessarily to automate an entire job.
It's to look inside the job and understand where different types of work should live.
There are things humans are exceptionally good at. Relationships, judgment, empathy, persuasion and navigating nuanced situations are difficult to reduce to a workflow.
There are other parts of work that depend much more heavily on consistency, monitoring, repetition and responding quickly when certain conditions change.
Historically, we've bundled all of that together and called it someone's job.
AI agents give businesses an opportunity to start unbundling it.
Now Think About the Recruiter's Job
This is where the conversation becomes particularly interesting for staffing.
Recruiting is fundamentally a relationship business.
A great recruiter understands what motivates someone. They build trust over time. They know how to position an opportunity, work through hesitation, negotiate and help a candidate make a decision.
We don't want less of that.
We want more of it.
But that's not everything a recruiter does.
A significant amount of their day is also spent searching systems, monitoring candidate activity, updating information, checking for matches, sending routine communication and trying to remember when hundreds of different people need attention.
Technology has helped make those tasks faster over the years, but we've mostly kept the same operating model:
The recruiter still has to make the work happen.
That's the assumption AI agents begin to challenge.
Stop Asking How AI Can Make Recruiters Faster
For the last several years, this has been one of the dominant questions in recruiting technology:
How can we make recruiters more productive?
It's a reasonable question.
But agentic AI gives staffing leaders an opportunity to ask a better one:
What work actually requires a recruiter?
If you were building a healthcare staffing agency from scratch today, knowing what technology is now capable of, would you design the recruiter's job exactly the way it exists today?
Would a recruiter need to manually initiate every routine candidate interaction?
Would they need to continually search a database to discover whether someone now matches an opportunity?
Would they need to remember every preference change, availability date and follow-up?
Would your most valuable relationship builders spend a significant portion of their day monitoring software?
Probably not.
Yet much of our staffing technology was built around exactly that model.
That's Why "Redesign" Matters
When Deloitte says leaders expect business processes to be redesigned around AI agents, that's much more significant than saying companies plan to automate more tasks.
Automation takes an existing process and makes parts of it faster.
Redesign asks whether the process should work that way in the first place.
That's the opportunity staffing leaders should be paying attention to.
Don't simply take your current recruiting workflow, put AI on top of it and call it transformation.
Start with the outcome.
Then decide where a recruiter creates the most value, where AI can take responsibility for defined work and where the two should interact.
That creates a very different operating model.
Healthcare Staffing Is Particularly Well Suited for This Shift
Healthcare staffing combines incredibly valuable human relationships with an enormous amount of changing information.
A candidate's fit for an opportunity can depend on specialty, licensing, location, shift, compensation, availability and preferences that may have changed since the last conversation.
Jobs also move quickly.
That means recruiters aren't simply managing people. They're constantly monitoring a changing relationship between candidate data and job data.
AI is exceptionally well suited for helping make sense of that complexity.
But there's an important caveat.
An AI agent can't operate intelligently if the information underneath it isn't accurate.
If candidate preferences are outdated, the agent is working with outdated preferences. If your database is filled with duplicates, dead contacts or incomplete profiles, adding more sophisticated AI doesn't magically make those problems disappear.
It can amplify them.
That's why the foundation for agentic AI isn't just the agent.
It's data, context and industry understanding.
Before an AI system can take meaningful action, it has to understand the environment in which it's acting.
In healthcare staffing, that requires more than a generic AI model understanding the word "nurse."
It requires understanding how healthcare staffing actually works.
From Software to Workforce
For decades, we've bought software for our workforce.
The ATS stores candidate information. The CRM helps organize relationships. Matching technology surfaces opportunities. Communication platforms help recruiters reach people.
But the human still sits at the center of those systems, moving the work from one step to the next.
Agentic AI introduces something fundamentally different.
We're beginning to move from technology that simply supports the workforce toward technology that can participate in completing defined pieces of the work.
That's why I believe the next phase of AI will be much bigger than another generation of software.
We're entering the AI Workforce era.
That doesn't mean replacing your human workforce with an artificial one.
It means intentionally designing a workforce where people and AI each spend more time doing the work they're best suited to do.
For staffing agencies, that could mean recruiters spending less of their day operating software and more of their day building the relationships that actually produce placements.
Start With the Work, Not the Technology
There's going to be no shortage of AI products sold to staffing agencies over the next few years.
Some will be transformative. Some will be useful. Some will simply add "AI" to an existing feature.
That's why your AI strategy shouldn't start with:
What AI tools should we buy?
Start with:
What work should our people actually be doing?
Look at how your recruiters spend their day. Identify where their expertise creates real value and where their time is being consumed by work that exists simply because technology previously required a human to move the process forward.
Then ask what your organization would look like if you could redesign that division of work today.
Because the companies that win this transition won't necessarily be the ones that buy the most AI.
They'll be the ones that figure out the right relationship between their people and their AI.
For staffing leaders, there's one question I'd start with:
If you were building your recruiting operation from scratch today, would you design it the same way?
We wouldn't.
And at Ember, we've spent a lot of time thinking about what we'd build instead.
More on that soon. 🔥