Marcus runs TA at an 80-person SaaS company. His open role: Senior Data Engineer. Three interview rounds, strong scores, a clean offer letter.

Day one, the new hire's camera stays off. By day three, the team notices the code doesn't match the interview. By day nine, nobody is sure who they actually hired.

The background check came back clean. It always does. It ran after the offer.

Marcus did everything right. That's the uncomfortable part.

Pre-employment screening software verifies the person. It was never built to question the story.

Why pre-employment screening software runs too late

You're checking IDs at the exit

Most screening starts after the offer. Identity. Criminal record. Employment dates.

That's a bouncer checking IDs on the way out.

By then you've already paid for three rounds, four calendars, and one hiring manager's patience. Marcus lost eleven interviewer-hours before anyone asked a hard question.

Run the numbers on his funnel: 40 applicants, 9 phone screens, 4 finalists, 1 offer. The check touched exactly one person. The other 39 never faced a hard question about their story.

Faster hiring didn't fix this. It only made the late check later.

The fakes arrive at the top of the funnel

GoodTime's 2026 hiring survey ranked fake and AI-assisted candidates as TA leaders' top worry. Above "we can't find qualified people."

Résumé written by AI. Interview answers whispered by AI. Sometimes a profile with no real person behind it.

Remote hiring made it easier. No office. No handshake. A face on a screen and a link to a repo.

Screening software that waits for the offer meets all of them too late.

Fraud isn't one thing

Three kinds of candidates walk into your pipeline.

• The fake: no real person.
• The coached: real person, borrowed answers.
• The stretcher: real person, padded story.

Post-offer screening might catch the first one. Maybe. The other two pass with flying colors.

Different problems. Same blind spot: nobody tested the story while testing it was still cheap.

Every round trusts the round before

Round two assumes round one checked.

Round three assumes round two did.

Nobody checked. Everyone assumed.

Marcus's three interviewers each wrote "strong signal" on the scorecard. Not one of them had verified a single claim.

It's a relay race of assumptions. Everyone holds the baton. Nobody opens it.

The 3-Gate Screen: move the doubt before the first interview

Marcus rebuilt his pipeline in a week. Same tools. Same team. Different order.

It works with a spreadsheet. Three gates, cheapest first.

Gate 1: Paper

Before any call, read the résumé against itself.

• Overlapping dates
• Titles that outrun tenure
• Skills with no job behind them
• Wins with no numbers

Four minutes per résumé is plenty. You're not investigating. You're reading carefully.

Timeline = story. If the story doesn't add up, mark it. Don't bin it.

A flag is a question for the call, not a verdict on the person. Some people have messy careers and real stories behind them.

Gate 2: Probe

In the first call, pick one project on the résumé. Ask three follow-ups only the person who did the work can answer.

"What broke?"

"What did you change afterward?"

"What would you never do again?"

Skip trivia. Ask about decisions. Trivia can be searched mid-call. Decisions can't.

Rehearsed answers are smooth. Real ones have scar tissue. It's the same trap now facing

AI video interview software

: a polished answer proves nothing by itself.

Gate 3: Proof

Verify identity and one piece of real work before the final round. Not after the offer.

• Live ID match
• A real reference call
• A small work sample

Cheap now. Expensive on day nine.

Keep it light. This isn't an interrogation. It's the same ask for every finalist, so nobody feels singled out.

Gates, not walls. Each one raises a question. A human answers it.

The math is short.

• Gate 1: about four minutes per résumé.
• Gate 2: three extra questions.
• Gate 3: one email.
• Skipping all three: eleven interviewer-hours, plus a day-nine surprise.

The Knoot Angle

Pre-employment screening software does one job well: confirming facts, late. The 3-Gate Screen works because it moves doubt earlier, when it's cheap.

That's where AI Screening in Knoot sits: Gate 1. It reads every candidate against the criteria from your JD, ranks the shortlist by match %, and explains why each person made it. It also surfaces risk flags: timeline overlaps, skill claims that don't match the experience around them, exaggerations. Think of it as a second pair of eyes that doesn't get tired at résumé number 300. The same reason

regulators now treat AI screening as high-risk

is why every flag comes with its reasoning.

Flags, not rejections. A flag means "look here," not "drop this person." You decide. Knoot doesn't verify identities or run background checks. Gates 2 and 3 stay with you and the tools you already trust.

What changes is the order. Not the tools, not the headcount, not the budget. By the time you spend an interview hour, the paper has already survived a hard look.

The AI reads the paper. You read the person.