Pipeline
Naming AI Tools in a Posting Versus Naming the Behavior
Describe what the person does with AI, not which products they use. A tool name in a job posting is a keyword, and applicants mirror keywords back, so the products listed in the requirements block come back on nearly every resume and tell you nothing about who uses them. Name a product only where nothing substitutes for it: a licensed clinical system, a proprietary stack, a workflow a regulator names by product. Everywhere else, name the artifact and what the person has to do to it.
The takeEverything written about tools versus skills is written for the candidate side, coaching people on how to list ChatGPT. The employer side inverts the stakes. A candidate listing a tool is making a claim about themselves. An employer listing a tool is publishing the answer key, and there is a measured gap in advertised salary attached to AI language in postings, so the incentive to copy the wording back is not hypothetical. Publish what the work needs and let the resumes differ.
Where Olive fits
Open a role and see what the work shows
Naming the behavior only helps if something further along observes it. Olive is one place that happens: a role-grounded assignment with an AI assistant that will do the whole thing if nobody stops it, returned as six findings a human reviewer writes from what the candidate actually did.
Rank your shortlistWhy a named tool stops being a signal
Because naming it teaches everyone the answer. A posting is read by applicants who want the job, and the requirements block is the most-copied part of it. List four products and you get four products back, on resumes written by people who used one of them twice and by people who use all four daily, in wording you can no longer tell apart. The line has removed the difference it was meant to reveal.
The incentive to copy is measurable, not theoretical. Lightcast reports that salaries in postings mentioning AI skills run 28% higher than in postings that do not, roughly $18,000 more a year, and that over half of postings requesting AI skills in 2024 were outside IT and computer science 1. That is a raw comparison of advertised salaries between two groups of postings rather than a like-for-like wage estimate, and AI-mentioning postings skew senior. It still tells a candidate exactly where the money is written, which is enough to shape what goes on a resume.
It gets worse where the tracking system picks the name up. A product listed in the requirements block tends to become a keyword filter or a knockout question downstream, so the line stops being descriptive and starts removing people, and it removes them on a word rather than on anything they can do.
So the product list ends up doing the opposite of its job. It was added to help sort the pile and it makes the pile more uniform. What survives it is anything a candidate cannot mirror without having done it: a described piece of work, a specific thing that went wrong, an account of what they threw away. What to do when a candidate lists six AI tools is the downstream version of this problem, once the mirroring has already happened.
When to name a product anyway
When nothing else does the job. Three cases justify a product name in a posting, and they share one property: substitution is not available to the new hire. A licensed system the role sits inside, such as a clinical documentation platform or a regulated filing tool. A proprietary internal stack a candidate could not have touched elsewhere. And a workflow a regulator or a client names by product in a contract.
In those cases the name is information rather than a keyword, because it changes what a candidate should expect on day one. It is also worth saying plainly whether prior experience with it is required or whether the team trains people into it, since a licensed system is usually the second even when the posting reads like the first.
One case that looks like an exception and is not: naming a tool to advertise that your team has access to it. That belongs in the description of the work or in the benefits, not in the qualifications, and it should say what a candidate gets rather than what they must already have. The same drafting decision by column, required or preferred, applies to a product name exactly as it does to a skill.
Write the behavior and the artifact instead
Name the thing produced and the person's relationship to it. A behavior line has three parts: the artifact, the assistance, and the accountability. Drafts the monthly variance commentary with an assistant, reconciles every figure against the source system, and can say which parts of the draft were discarded. A candidate reads that and knows what Tuesday looks like. A screener reads it and knows what to ask about.
Four worked swaps, each replacing a product name with what the product was standing in for:
- 'Proficiency with ChatGPT and Claude' becomes 'drafts first-pass client memos with an assistant and is accountable for every citation in them'.
- 'Experience with GitHub Copilot' becomes 'ships code written with an assistant and can explain what the tests cover and what they do not'.
- 'Familiar with Midjourney' becomes 'produces concept boards with a generation tool and can say which references were rejected and why'.
- 'Prompt engineering skills' becomes 'gets useful output from an assistant on unfamiliar problems and can show how the ask was narrowed'.
Each swap survives a version bump, and each one gives an interviewer something to open. The wording that fails is the hybrid: naming a tool and then adding proficiency, which asks for a keyword and a self-rating at once. What good AI use actually looks like in an interview is the reference to write these against.
How long does a tool name stay accurate?
Nobody publishes a number, so the answer that helps is a re-reading interval. What is measured is the language around the names: Indeed Hiring Lab put AI-related postings at 6.3% of US postings in its August 2026 snapshot, well past a prior peak of 3.3% in 2022 2. That counts postings mentioning AI and does not track which products they name, so it dates the vocabulary while saying nothing about the shelf life of any one product.
The underlying work moves more slowly, which is exactly why a behavior line lasts. Indeed's GenAI Skill Transformation Index rated almost 2,900 work skills and mapped them onto more than 53.5 million postings, finding 40% of skills minimally transformed, 19% assisted, 40% hybrid and 1% fully transformed 3. Those ratings come from language models scoring skills rather than from watching work being done, so read them as a judgment about capability. Hybrid transformation explicitly means human oversight is still needed, and it is not a synonym for automatable.
The practical consequence for a recruiter is a maintenance rule. Whatever a posting names by product, somebody has to own re-reading it every couple of quarters. Whatever it names as a behavior can sit for a year without embarrassing anyone. Given how many roles a small team keeps open at once, that difference decides whether the postings stay honest or quietly rot in the requisition library.
A related decision worth making at the same time: whether the posting should describe how you evaluate, which ages on the same clock and is read by the same person.
Common questions
Will dropping tool names hurt how the posting matches on job boards?
Somewhat, and the trade is usually worth it. Board matching reads the whole posting, so a behavior line mentioning an AI assistant still surfaces on general AI searches, though it will miss a search for one exact product. If discovery is the binding problem, put the product names in the description of the work rather than in the qualifications: they remain searchable, and nobody reads them as a gate they must clear.
What if the team genuinely does standardise on one tool?
Then say so in the duties, and say whether prior experience is expected. A sentence like 'the team drafts in one assistant, which you will be set up with in week one' tells a candidate what to expect and asks nothing of their history. Standardising on a product is a fact about your environment, not a qualification, unless the system is licensed or restricted enough that outside experience genuinely cannot be substituted.
Do behavior lines make the posting longer?
By a few words per line, and shorter overall in practice, because one behavior line usually replaces three or four product bullets that were saying the same thing. The bigger change is where the words sit. A qualifications block shrinks, a duties section grows, and the posting starts reading as a description of a job rather than as a checklist somebody has to score themselves against.
Should the posting say which AI tools candidates may use when applying?
That is a separate decision and worth making explicitly, because silence gets read as permission by some applicants and as prohibition by others. If there is a rule for the application or a later exercise, state it once, in plain terms, where the instructions for that stage live. Do not bury it in the qualifications block, where it competes with the description of the job and reaches people who have not applied yet.
How do I get a hiring manager to drop their tool list?
Put the question back to them: what would a resume listing every tool on the list actually tell you. The usual answer, once said out loud, is nothing they could act on. Then ask what they would want to hear in a first conversation, and write that down: it is almost always a behavior, phrased as something the person did to a piece of work. The list tends to fall away on its own once the alternative exists on paper.
References
- 1. Beyond the Buzz: Developing the AI Skills Employers Actually Need lightcast.io Supports the claim that postings mentioning AI skills advertise higher salaries, which is the incentive candidates have to mirror posting language.
- 2. US Labor Market Snapshot: August 2026 hiringlab.indeed.com Supports the dated share of US postings carrying AI-related language and its rise past the earlier peak.
- 3. AI at Work Report 2025: How GenAI is Rewiring the DNA of Jobs hiringlab.indeed.com Supports the claim that most work skills are changed rather than replaced, which is why a behavior line outlasts a product name.
3 sources, numbered by first appearance. How Olive sources claims
General guidance for hiring teams. What works at one company and one volume may not transfer to yours.
Olive assesses how a person works with AI. It does not detect AI-written documents, and it never produces a score, a ranking, or a match percentage for a person. Candidates read the same report the employer reads.