Recruiting for healthcare is a category of its own. One in three new jobs in the US is in healthcare (Bureau of Labor Statistics). Turnover sits at up to 50%. The average time from req open to start is 82 days. And the average healthcare recruiter is handling 143 hires a year while reaching fewer than 3-4% of applicants in high-volume roles.

The math doesn't work – and scaling recruiting teams isn't the answer.

That's how Joe Gage framed the challenge at ASHHRA 2026, when he took the stage with Yaniv Shimoni, Co-Founder of Take2 AI. As Chief Human Resources Officer and Chief Administrative Officer at Bon Secours Mercy Health – a 60,000-person health system spanning 14 states – Joe has watched every flavor of HR technology come and go. At this year's conference, he shared the questions he actually asks before adopting any agentic AI solution in talent acquisition.

If you're evaluating vendors right now, this framework is worth your time.

See why healthcare's leading talent acquisition teams recruit with Take2. Book a free demo →

1. Is it actually autonomous?

This is the first question Joe raised, and it's the one most vendors would rather you not ask. A lot of platforms use the word "agent" – but if the system requires constant prompting or human intervention to complete tasks, it isn't really autonomous. It's just a more expensive workflow tool.

The distinction matters operationally. An agent that needs hand-holding defeats the purpose. You're not freeing up recruiter time if someone still has to babysit the process.

"A lot of platforms will use the word 'agent.' If it needs hand-holding, it's just a more expensive workflow tool." — Joe Gage – CHRO and CAO at Bon Secours Mercy Health

2. Does it speak the language of healthcare?

Healthcare is genuinely different. When Joe joined the industry in 2003, someone had to hand him a glossary just so he could survive a meeting. LLMs are not created equal, and an AI system trained on generic data will miss nuance that matters – clinical role requirements, credentialing terminology, role-specific screening criteria.

An agent that interviews a phlebotomist candidate needs to know what that conversation should sound like. One that doesn't will produce poor candidate experiences and worse data.

"LLMs are not created equal. An agent that interviews a phlebotomist candidate needs to know what that conversation should sound like." — Joe Gage

3. Does it integrate with your ATS?

If the system doesn't write back to your ATS, you're giving your recruiters one more screen to manage. That's not a feature – it's friction. Joe was direct: full bidirectional integration isn't optional. It's the difference between augmenting your existing process and disrupting it.

"If the system doesn't write back to your ATS, you're giving your recruiters one more screen to manage. That's not a feature – that's friction." — Joe Gage

4. Is it staying in its lane on decisions?

This is where legal and compliance teams get nervous – and rightly so. Joe, a former labor and employment attorney, was unambiguous: AI is not ready to make selection decisions. What it should be doing is collecting information, not evaluating it. Screening a candidate, scheduling them, verifying credentials – yes. Deciding who gets hired – no.

Any vendor that blurs this line is a liability, not a solution. For healthcare organizations especially, the legal exposure from an AI that crosses that line isn't hypothetical.

5. Is it a single-agent or multi-agent system?

Joe's view here was clear: the single-agent era is over. A solution that automates one step of your workflow – just scheduling, or just screening – isn't going to move the needle on time-to-fill. The impact comes from stringing together a sequence: sourcing, screening, scheduling, credential verification. That's where you get 70-80% workflow automation, not 15%.

"The single-agent era is over. The impact comes from stringing together a sequence – that's where you get 70–80% workflow automation, not 15%." — Joe Gage

6. What does it actually cost – and how is it priced?

Pricing models matter – and the question isn't just what you pay, but how predictable it is at scale. In high-volume healthcare environments where you're processing thousands of applicants per week, costs can compound quickly. Joe's framing: think of it less like buying software and more like adding two or three recruiting coordinators. If the price roughly matches that, you're in a reasonable range. If it's dramatically more, you need a clear ROI case before you sign.

7. How much does it disrupt your existing workflow?

The best agentic solutions bolt into what you already have rather than requiring a rebuild from scratch. Joe contrasted this with a Workday ATS implementation – a major overhaul that took significant time and created real organizational anxiety. A well-designed agentic solution should run in parallel with your existing team, not replace your process before you're ready.

At Bon Secours Mercy Health, the framing Joe used internally was simple: it gives recruiters the tools to do the job they actually can't get done today. The average healthcare recruiter handles 143 hires per year, and roughly 1 in 10 of those hires is gone within 90 days. That's not a recruiting failure – it's a data-quality failure driven by exhaustion and volume. AI that handles information collection means recruiters are still sharp when it's time to make the calls that require judgment.

The bottom line

Healthcare TA leaders are being pitched agentic AI from every direction right now. The vendors using the word "agent" are not all building the same thing.

Joe's seven questions – autonomy, healthcare fluency, ATS integration, decision-making boundaries, multi-agent depth, pricing structure, and workflow compatibility – give you a checklist that cuts through the noise fast. Start here and you'll disqualify the wrong vendors in the first conversation.

The technology is real and it works. The due diligence is yours to do.

How Take2 answers every one of these questions

Take2 was built for this checklist. Its agents run fully autonomously across the full recruiting workflow: sourcing, screening, scheduling, and credential verification. Every agent is trained on healthcare-specific language and role types, so candidate conversations sound right and produce data your recruiters can actually use. It integrates bidirectionally with every major ATS system, pushing scorecards and recordings directly back to the applicant record. The result: healthcare organizations using Take2 are cutting time-to-hire in half and significantly reducing sourcing costs.

See how Take2's AI agents can transform your recruiting operation. Book a 30-minute call. We'll walk through projected impact based on your role types, requisition volume, and recruiter capacity.

WATCH: Joe Gage on why health systems need agentic AI in recruiting