I think skills-based hiring will keep disappointing until the skills layer in HR systems records how a skill was shown. Two releases this week point at the same gap from opposite ends. One measures what AI is taking over at the level of tasks. The other checks whether a worker can still do the task.

Credentials say someone passed a course

Most skills data an HR team holds is a credential, a self-reported tag or a manager’s rating. Interplay Learning, which sells simulation-based training for the skilled trades, launched a product on October 15 aimed at the space between those records and the job. In its release, the company says employers have “limited ways to verify hands-on ability before putting someone on the job.”

The strongest line in the release comes from a customer. Matt Ellenberg, director of training at Southern Home Services, said: “You can have your OSHA, EPA and NATE certifications and still not be able to perform the job in the real world.” A training director saying that about three well-known certifications is a useful data point for anyone building a skills taxonomy, because a taxonomy that treats each certification as proof of the skill underneath it will overstate what the workforce can do.

Advertisement

HRTech Your brand belongs here. Reach the decision-makers who read HRTech every day. Premium placements across the site and newsletter. Advertise with us

The product itself is modest. Interplay says the first catalog holds 20 assessments across commercial HVAC, facilities management, manufacturing, data centers and critical infrastructure, including Using a Digital Multimeter and Industrial Electrical Lockout/Tagout Safety Procedure. Each one produces a score plus step-level detail on missed steps, sequence and decisions. The company positions the results for use before dispatch, contractor authorization, new-hire placement or promotion decisions.

Task-level data is arriving from the other side

On October 14 the American Staffing Association released an AI Impact Index that draws on Anthropic data and the U.S. Department of Labor’s O*NET database. It examines the share of each task in an occupation that AI automates and augments, then weights each task by its importance to the job. The first results list software quality assurance analysts and testers, writers and authors, and data entry keyers as the occupations most affected, and 183 occupations with no automation or augmentation detected as of yet.

Noah Yosif, chief economist at the association, said staffing and recruiting professionals “can use the ASA Impact Index to anticipate client needs and position themselves as true workforce solutions partners.” The index is built for staffing firms, but the method matters to every HR team. Once demand is described task by task, a skills profile that lists a job title and a certificate is too coarse to match against it.

What I would change in the skills layer

I would add three fields to every skill record an HR system stores: how the skill was shown (course, certificate, assessment, observed work), when, and by whom. A skill with no method on file stays visible but carries less weight in matching, internal mobility and contractor authorization. HR teams that already use a skills-inference tool should ask the vendor how the tool treats a skill that was inferred from a job title compared with one that was tested. Our earlier piece on skills-data contracts argued that buyers should know who owns this data when a vendor changes hands. The method field belongs in the same conversation, because it decides whether the data is worth owning.

Newsletter

Get the week's best tech coverage.

Free. Read by thousands of HR, tech, and business leaders.

The market is already moving toward verification. We covered Pearson’s agreement to acquire Workera, a deal framed around verifying workforce skills. Interplay’s launch is the same idea applied to hands-on trades work.

The counter-argument

The fair objection is that a simulation of 20 tasks cannot stand in for a trade, and the release cites no study showing that scores predict performance on the job. Certifications also carry legal weight in regulated work, and an employer cannot drop them because a simulator exists. I accept both points. The release itself does not claim a replacement. Doug Donovan, CEO and founder of Interplay Learning, describes the product as giving operations and safety leaders “objective evidence of demonstrated performance,” which adds to the file without replacing anything in it. A file with a certificate and a recorded assessment says more than a file with a certificate alone.

The validity question is also one buyers can put to the vendor before signing. Ask for the evidence that scores track field results, ask how often assessments are updated when equipment changes, and ask whether results export to your HRIS in a form that keeps the step-level detail. If the answers are thin, the tool still works as a training aid and should be labelled that way in the skills record.

Sources: Interplay Learning, “Interplay Learning Launches Skill Verification to Demonstrate Workforce Readiness”; American Staffing Association, “New ASA AI Impact Index Reveals the Jobs Most Impacted by AI”