Founder Spotlight: Bennett Kim (ZNest)

Bennett Kim is the Co-Founder and CEO of ZNest, an AI-powered hiring and credentialing automation platform built specifically for senior living. ZNest helps operators reduce applicant drop-off, streamline onboarding, and ensure regulatory compliance—all without increasing staff burden. At Senior Living 100’s Startup Pitch Competition, I was honored to join the panel of Judges that selected ZNest as the Judges’ Pick for its practical, immediate impact on one of the most pressing challenges in senior living: staffing. With a background in building high-performance systems and a deep appreciation for operational realities, Bennett is on a mission to bring modern workforce infrastructure to an aging sector in need of innovation.

— Sarah Richard

P.S. - If you’re an AgeTech founder, investor, or operator based in the NYC area, please consider joining us at FirstGen Founders Powering the Care Economy

Could you tell us more about your path to building ZNest? 

BK: My first experience with senior living was over 20 years ago when I was with the private equity firm, Apollo.  Apollo back then owned one of the largest assisted living operators in the country called Summerville.  Over the next couple decades as an investor and developer, I became aware of many issues within the industry.  The two things that I wanted to help address were the difficulty for families to find the right fit in terms of senior housing options and the affordability of middle-income assisted living.  I started off focusing on the former but then pivoted to the latter after getting a better understanding of potential solutions.  Our team and I realized how much AI could play a role in making senior housing more affordable for the average American family.  When the tech world talked about how overhead could be streamlined dramatically using AI, we knew this path was how we could have a bigger impact in changing the cost structure of an entire industry. 

Staffing is one of the biggest pain points in senior living. What inefficiencies did you observe in the hiring process, and how does ZNest streamline those workflows for operators?

BK: There’s a fundamental difference between how we used to think about solutions when looking through the lens of developing software versus how we now think about solutions using AI.  When we thought about things from a software perspective, we tried to identify the most painful workflows.  Why would someone buy software from us?  It’s either because we’re solving something so painful that the customer is either willing to switch from whatever it was using before or that we were able to identify something that nobody else thought to fix.  A third option is that the pain comes from how expensive the current solution is and we can then pursue a lower cost option.  AI is different.  We can now do things faster and with fewer mistakes at a fraction of human labor costs.  Now the question is, “Where are the workflows that are manual and repetitive but necessary?”  Any workflow with those three characteristics makes AI worthy of consideration… even if it is not painful.  

We saw that many functional areas had these characteristics, but in human resources and, specifically, within recruiting and onboarding, we saw an opportunity to not only make workflows more efficient but also to help address the labor shortage.  We saw that all the onboarding requirements that help make senior living a safer industry were also making jobs less desirable than competitive front-line roles in other industries.  By addressing the inefficiencies in the hiring process, we hope to have a larger impact on the labor pool.  

Operators often cite credentialing and compliance as time-consuming and error-prone. How does ZNest help reduce risk and support regulatory accuracy without increasing admin workload?

BK: There is no reason why an educated and talented HR manager should be reading the contents of a license or certification, typing them into a spreadsheet, checking websites to see that the contents match, and monitoring that spreadsheet continually.  Those tasks require no judgment.  There are also so many points that humans can make errors throughout that process, and the consequences are huge, sometimes existential.   

For now, we are focusing on areas that do not require judgment, simply manual tasks.  Let me clarify, though, the outcome is deterministic but the process to get to the outcome requires a little bit of judgment.  For example, we’ve trained the AI models to know that it must verify any type of credentials.  The process in which it knows the job position, state, and governing bodies is based on AI learning on its own.  The AI model extracts data from the credentials using computer vision and understands what they mean.  We realize the gravity of making sure everything is accurate which is why we use a multi-agent network.  This network is essentially a team of different AI foundational models from different companies that confer with each other and only arrive at an answer when they all agree.  If there is anything questionable, we make sure to alert humans.  

What outcomes are you tracking to measure ZNest’s impact, and what have you seen so far from early partners or pilot communities?

BK: For now, we only have enough data for the front end of our tool which is to handhold job applicants through the onboarding requirements like a background check, TB test, and drug screen using conversational AI texting.  One beta tester told us that prior to using our tool 15-20% of job applicants missed these appointments.  After using our tool, not one has missed so far.  We’ve also reduced time-to-hire by two days, which can have a huge impact in terms of making sure candidates show up for their first day of work.  We’re currently testing the credential verification and management features and should have more metrics soon. 

Looking ahead, what’s next on the ZNest roadmap? Are there other aspects of workforce management or operations you’re excited to tackle?

BK: Because we are not creating software, we don’t have to necessarily take a linear path in terms of the roadmap.  In other words, many software companies look at a roadmap and think about how to go deeper or wider from where they started because… it’s software.  It has to be connected somehow.  We look at solutions as a series of AI agents.  The agents automate workflows.  The agents may work with whatever software operators currently use.  In fact, there may be cases where the agents manage the software.  Where we are heading next is in risk management.  There is logic to our roadmap, but we’re going to keep that on the down-low for now.