AI fluency is now a line item on the scorecard. Hiring managers ask for it, leadership expects it, and candidates list it on every resume. The problem is that almost nobody knows how to score it in an interview, so it gets waved through on vibes: a confident answer about ChatGPT, a few buzzwords, and a yes.

Zapier was the first company to do something about that publicly. Its CEO, Wade Foster, published the AI fluency rubric Zapier uses in hiring, with named levels that run from resistant to transformative, and made passing it a real bar for the roles they fill. It turned a fuzzy buzzword into something you can grade.

Most talent teams want the same thing and stall at the same place: what are the levels, what do they sound like in a real conversation, and how do you score one without rewarding the candidate who's just good at talking about AI? This is the rubric, the questions that surface each level, and how to score it from what was said rather than from a gut feel two days later.

Why AI fluency is suddenly on the scorecard

Two roles can have the same title and the same years of experience, and one of them now ships three times the output because of how they work with AI. That gap is widening fast enough that teams have stopped treating AI fluency as a nice-to-have and started treating it as a core competency, the same way they treat communication or ownership.

Some teams have made it explicit. They probe it in the process so they don't accidentally hire someone who will resist the way the rest of the team already works.

The catch is that wanting to assess AI fluency and being able to assess it are different things. Ask a room of recruiters how they score it and you'll get a shrug, or a story about a candidate who sounded impressive and then couldn't do the work. The honest read across the field is that the requirement has arrived well ahead of the method:

So the job isn't to decide whether AI fluency matters. That's settled. The job is to define what you're looking for, in levels you can tell apart, and then build the interview to surface them. That starts with a shared definition.

What AI fluency is: the four levels

AI fluency is the ability to get meaningfully better results by working with AI, and to know when not to trust it. It isn't whether someone has used a chatbot. It's how deeply AI is woven into how they work, and how good their judgment is about where it helps and where it quietly gets things wrong.

Borrowing the shape of Zapier's public rubric and tuning it for an interview, four levels do the work. Read across the row to score a candidate: what the level looks like, what you'll hear when someone is at it, and how to weigh it.

The line that matters most sits between Capable and Adoptive, and it isn't about tools. It's judgment: whether the person knows where AI fails and works around it on purpose. A fluent operator treats the model as confident and often wrong, and builds that assumption into how they use it.

That instinct, knowing the tool tilts a certain way and designing for it, is exactly what you're listening for in a candidate. Someone who only ever describes AI's wins is usually a level below someone who can tell you, in detail, the last time it let them down and what they did about it.

How to interview for each level

The rubric only works if your questions surface real behavior instead of rehearsed opinions. Three moves do most of that work.

1. Ask for one real example. "Walk me through the last thing you used AI for, end to end." A lived workflow exposes the level; a hypothetical just rewards whoever talks about AI most fluently.
2. Follow the work, not the words. Probe the steps: what did they ask, what came back, what was wrong, what did they change? Capable stops at "it gave me a draft." Adoptive and above can narrate the loop.
3. Test for judgment. "When has AI been confidently wrong for you, and how did you catch it?" The answer separates someone who trusts the output from someone who supervises it.

Notice what these have in common: none of them ask whether the candidate likes AI, and none can be passed with a buzzword. They make the person show you a workflow, which is the only place fluency really lives. Define what a strong, average, and weak answer looks like for each before the interview, so a yes means the same thing whoever gave it.

Score it from the interview, not from memory

A rubric is only as good as the evidence behind each score. The usual failure is that the interviewer asks a sharp AI question, hears a great answer, and then writes the scorecard two days later from a fuzzy memory of it. The level gets rounded up, and the rubric becomes decoration.

This is where the capture layer earns its place. Because Notetaker captures every spoken word, the candidate's actual example, the workflow they described and the failure they caught, is on the record instead of in your head. You score against what was said, not what you half-remember.

From there, the score writes itself against your template. Metaview drafts the scorecard from the conversation using the rubric you set, so the AI fluency level is filled in from what the candidate described, ready for you to confirm or adjust.

Then Reports closes the loop across the whole pipeline. You can see whether interviewers actually assessed AI fluency where they were supposed to, how the level breaks down by team, and where a strong yes from one interviewer means something different from another's. That's how a rubric stays consistent instead of drifting role by role.

None of this replaces your judgment. It gives the judgment something solid to stand on. And the reason it's worth the effort is that the teams treating AI as core to how they hire are the ones pulling ahead on the numbers that matter:

What this means for your team

Pick one role you're hiring for right now and decide, out loud with the hiring manager, which level it genuinely needs. A junior analyst probably needs Capable. A role you're betting the function on probably needs Adoptive or higher. Writing that line down is most of the work, and it's the part teams keep skipping.

Then put the rubric where the interview happens. Add the AI fluency questions to that loop, build the bar into your question bank and your scorecard templates, and let the capture layer record the answers so the score is evidence, not recall. Connect it through native integrations and keep the ATS you already run. If you want the groundwork first, our writeups on great interviewers, quality of hire, and interview quality set it up, Reports keeps the bar honest across the team, and pricing shows what it costs.

AI fluency stopped being optional faster than most hiring processes adapted. You don't need a perfect science to keep up. You need a shared definition, a few questions that surface it, and a way to score it that doesn't depend on who was in the room. Zapier proved a rubric beats a vibe. This is how you run yours.