In this section I review one AI-powered application and demonstrate how it can be used to create new value.
Recruiting at any scale is a translation problem. The recruiter has a job description, with words on it. The candidate has a career, with words on it. The match between the two has to be made by a human who understands what both sets of words actually mean. There are usually too many candidates for the recruiter to read carefully, and too many jobs for any single candidate to evaluate on their own. The whole industry runs on shortcuts: keyword filtering, alma-mater bias, prior-job-title pattern matching, and network referrals. AI is supposed to fix this.
I have been working with a couple of recruiting firms over the last year, helping them think through where AI fits in their operations. I have not used Findem myself, but I have been close enough to the category to recognize this story when I heard it. RecruitMilitary, the largest veteran-focused recruiting business in the United States, is the cleanest published example I know of what good looks like when AI lands in the recruiting layer. It is also a clean example of Mandate 1 from this issue's HR dual mandate framework, transforming HR itself, with measurable outcomes.
A quick word on RecruitMilitary
Some context. RecruitMilitary is veteran-owned and employee-owned. CEO Tim Best is a former 160th SOAR helicopter pilot. They run more than 100 career fairs a year and operate the world's largest veteran-focused job board. In 2025 they generated 600,000 connections between employers and military-community job seekers. That is a lot of people moved across an industry famous for poor matching between veterans and the corporate roles they could excel in.
Six years ago, RecruitMilitary's CTO Mike Frankham started betting on skills-based hiring. He did not see Gen AI coming, but the directional bet was right and the timeline got compressed. Mike spent years evaluating veteran-focused matching technologies and was frustrated none of them understood the employer side. In 2024 he met Heather and the team at Findem at HR Tech, came back to Tim, and said: "I think I've found the partners." In 2025 they launched Veteran Talent Source, a Findem Co-Pilot variant.
What Findem actually does
It is a small mental shift with large operational consequences. A resume is a two-dimensional summary: jobs, dates, schools, titles. A job description is also two-dimensional: required skills, nice-to-haves, years of experience. Findem's view is three-dimensional. People are decomposed into hundreds of thousands of skills and attributes inferred from their full digital footprint. Jobs are decomposed into hundreds of thousands of skill attributes the recruiter could never tag manually.
Tim Best's framing on his recent What Works podcast episode with Kathy Andres is worth quoting directly:
"It gives you a view of people that's not two-dimensional like a resume. And a view of jobs is not two-dimensional like job descriptions. It recognizes that people are a mix of literally hundreds of thousands of skills and attributes."
The recruiter workflow is concrete and worth picturing. The recruiter clicks a job tile. The AI decomposes the description into more attributes than a human could ever select manually. The system returns a top-50 ranked match list with confidence scores. The recruiter can ask the AI to explain why a particular candidate was surfaced. Tim's example: "Tell me what a Navy machinist mate does and why they fit this role." The AI answers with thirty years of RecruitMilitary's domain data behind it. Then the recruiter triggers personalized outreach. The AI generates customized messages per candidate. What used to take days takes a few button pushes.
What the numbers actually say
The first-year metrics are striking. Personalized campaigns hit email open rates of 80 to 90 percent and click-through rates five times the industry norm. A typical recruiter campaign sees open rates in the 20 to 30 percent range and click-throughs in the low single digits. Five times that completely changes the unit economics of a recruiting team.
The most interesting line in Tim's interview is that the default fear about AI in recruiting is that it would depersonalize the candidate experience. Tim's data says the opposite:
"What I've seen is the exact opposite. These personalized campaigns will produce better and more personalized emails. And I think it actually produces a better experience."
That is what happens when AI does the personalization at scale that no human team can match. Bersin's Kathy Andres calls the human at the center of this kind of work the super worker, which I think is exactly the right frame. A recruiter armed with a tool like Findem is doing more, with more empathy, for more candidates than an unaided team possibly could.
A counter-voice from TechWolf
Now, the counter-voice. Mikaël Wornoo runs TechWolf, a Belgian competitor in the same skills-data-layer category, backed by SAP, ServiceNow, and Workday simultaneously. HSBC (200,000 employees) is both a customer and an investor. Wornoo agrees with the direction Findem and RecruitMilitary represent, but he is publicly more measured about the timeline:
"Right after Covid, a skills-based organization was going to be this completely new way of thinking about talent management. Well, that didn't really happen. It is not this entirely new paradigm that is going to change the world, but it is a much better way of running the known talent processes."
Worth holding both views. Tim Best is right that the technology is now ready to deliver real personalization at scale. Wornoo is right that most enterprise rollouts will fail unless the underlying skills data is good enough, which is the data-stewardship responsibility from this issue's leadership post on Earth's Best Employer showing up again. The category is real. The hype around it has consistently outrun the data quality.
Honest limitations
Findem and TechWolf are both enterprise sales motions, so you cannot try them yourself the way you can try Claude or ChatGPT. The platforms only deliver if the skills data, both candidate-side and job-side, is rich enough to feed the model. Neither is a substitute for the recruiter's judgment on the people who pass the surfacing step.
This is also where the skills-based organization concept I cover in this issue's terminology section becomes load-bearing. A skills-based recruiting tool sitting on top of a wishlist instead of a real skills graph will produce theatre, not results.
Your action step
If you run an internal recruiting team, audit your current recruiter workflow this week. How long does it take to go from "I have a role to fill" to "the first qualified candidate has received a personalized message"? That cycle time, and the open-rate and click-through metrics on outbound campaigns, is your before-baseline. Bring it to your next vendor conversation. If the platform you're evaluating cannot move those numbers measurably, the data underneath is the problem, not the tool.
If you're a senior leader sponsoring an HR transformation, this is also the easiest place to put a Mandate 1 win on the board, the kind of measurable, operational outcome that makes the room ready for the harder Mandate 2 conversations.
If you'd like to think through how skills-based recruiting fits into your AI transformation, or want me to run a working session with your HR and recruiting leaders, I'd love to help.
Sources
- Tim Best, RecruitMilitary, What Works podcast with Kathy Andres
- Mike Frankham, CTO, RecruitMilitary
- Mikaël Wornoo, TechWolf, Future Ready Leadership podcast
- Findem and TechWolf product overviews
- Josh Bersin / Kathy Andres on the super worker concept