Recruiters on AI: Boolean in 30 Seconds, Deepfakes in the Interview

Three veteran recruiters share where AI tools actually help, where they fail, and why the human element isn’t going anywhere.

AI is writing Boolean strings in seconds. It’s also fabricating candidate answers and, in some cases, candidates themselves.

Recruiters are caught in the middle—using the tech to speed up the grunt work while guarding against the scams it enables. Gloria Dallas (Snappr), Brandon Jeffs (building the Talent Machine), and Arielle Salman-Bailey (Warner Bros. Discovery) joined Wellfound’s Ask a Recruiter for ‘Recruiting Operations–What works and what doesn’t’ to talk about what’s hype, what’s useful, and what’s just plain strange.

🎥 Watch the full session on YouTube

AI Writes a Boolean Fast—But It Also Lies to You

All three panelists use AI to take the edge off tedious tasks: note-taking, resume ranking, Boolean string creation.

“A good Boolean string could take me 15 minutes to write thoughtfully,” Salman-Bailey said. “With AI, I can get it in 30 seconds and adjust if needed.”

Dallas uses it to refine job descriptions and filter resumes. But she’s learned to keep one eye on the output:

“I had an AI note-taker completely fabricate what a candidate said. When I asked why, it had no response. The AI just lied.”

Jeffs was blunt:

“There’s a need for a human in the loop. Someone’s gonna have to press play on automation. We don’t have solutions yet that are completely automated without a human.”

Where AI Helps—and Where It Doesn’t

Dallas summed up the trade-off:

“AI tends to know about more startups than I do. But sometimes I’ll say I need ten years of product management experience, and it gives me project managers instead.”

Useful

• Interview transcription — keeps hiring managers accountable and gives recruiters a record of what was actually said.
• Resume screening as a first pass — AI can stack-rank quickly, but humans still need to review the short list.
• Boolean and X-ray string generation — what used to take 15 minutes now takes 30 seconds and a tweak.
• Message sequencing for drip campaigns — AI drafts six-touch outreach so recruiters can focus on tailoring.
• Data cleanup for hiring manager decks — turns messy inputs into presentable insights fast.
• Market mapping — surfacing startups or candidate pools a human might not have time to uncover.

Not so useful

• Letting AI run an entire process end-to-end — note-takers fabricate answers, resume screeners confuse “product” with “project,” and sourcing filters mis-rank obvious fits.
• Relying on AI to judge cultural fit or nuanced human judgment — still squarely a people job.
• Over-filtering — tools that slice too aggressively often cut out great candidates.
• Generic, AI-generated outreach — candidates spot templated messaging a mile away.
• Video resumes — panelists said they mostly waste time and put candidates off, especially for technical or leadership roles.

“The vendors that win will be the ones with great first-party data,” Jeffs said during the panel. “A lot of tools scrape from the web or third-party sources, and that puts companies at risk.”

That’s why recruiters are leaning on platforms built from trusted, opt-in candidate data instead of scraped lists. Wellfound AI is one example—it pulls from millions of profiles where candidates have shared their information directly, so the sourcing results start closer to reality.

When Candidates Show Up Wearing Someone Else’s Face

The wildest stories weren’t about recruiters using AI—they were about candidates abusing it.
Salman-Bailey explained:

“We’ve had people interview virtually with deepfake filters of someone else’s face. Then they show up to the office and surprise—they’re not who they said they were.”

Dallas described catching a candidate copying AI-generated code mid-interview. Others shared experiences with candidates who literally forgot which fake name they were using.

How recruiters are fighting back

• Ask situational questions tied to the candidate’s own history. Salman-Bailey screens for details that only the real candidate would know. “If you’re reading off ChatGPT, the gaps show up fast,” she said.
• Set ground rules up front. Dallas emails candidates ahead of time to clarify when AI tools are fine (like looking something up mid-interview) and when they’re not. “If we notice you using it without permission, that’s a red flag.”
• Use transcripts as a safety net. Jeffs pointed out that transcription isn’t just about notes—it’s evidence. If a hiring manager later claims a candidate “wasn’t a fit,” the transcript shows who actually talked and what was said.
• Experiment with higher-friction formats. From flying candidates in for a day to piloting “clean room” style testing (phones and laptops left outside, work done under supervision), recruiters are trying ways to make cheating harder.

Deepfakes, fake names, AI-written answers—none of it makes someone better at the job. But apparently it makes them bold enough to try.

The Future Belongs to Recruiters Who Stay Human

For all the AI hype, the panel kept circling back to the same point: the best recruiters are the ones who can build trust and place people strategically.

Jeffs put it sharply:

“If you don’t want to be replaced, get really good at recruiting. Get really good at communicating strategically with hiring managers. There’s a culling in our industry, but it’s supply and demand.”

Dallas sees recruiting getting more personality-driven:

“We’ll have humans who are people-forward, maybe not doing as much data tracking. Really engaging people who can convince a candidate to leave their fancy big company to work at a startup.”

And Salman-Bailey reminded everyone that the basics don’t change:

“Even at a tech company, people want to feel warm and welcomed. I still have to be here 40 hours a week—I want to know the people I’m working with are still people, not robots.”

Tools Recruiters Actually Rate

• For high-volume screening: Boolean strings, Juicebox AI, Endorsed AI

• For relationship building: LinkedIn groups, industry-specific Facebook communities, old-fashioned networking

• For contact info: ContactOut, SeekOut

• For referrals: Internal programs that turn employees into mini-recruiters

• For sourcing at scale: Recruiters are shifting toward platforms built on first-party candidate data instead of scraped lists. Jeffs noted this is where vendors will sink or swim. That’s exactly what Wellfound AI offers—millions of opt-in profiles rather than patchy third-party scraping.

Salman-Bailey explains:

“Your network is sometimes your best tool. I stay engaged in LinkedIn groups and Facebook communities. When I need a specific role, people recognize me as a trusted source.”

Practical Takeaways

• Use AI for research and first passes—but review results manually
• Track data to flag bias in “cultural fit” rejections
• Add application questions to slow down spammy applies
• Leverage interview transcription for accountability and training
• Keep investing in real relationships—AI can’t fake trust (yet)

The tools will change again next quarter. The scams too. The only constant is that recruiters will be figuring it out live. The ones who last are the ones who adapt — the ones who can persuade someone to make a leap, and still feel human in a process that’s getting progressively more automated.

This article draws from Wellfound’s Ask a Recruiter live Q&A with Gloria Dallas, Brandon Jeffs, and Arielle Salman-Bailey. For more conversations like this, check out the full playlist here.