## **Key Takeaways**

1. At 10,000 applications per week, Vanta's recruiters were buried in manual review. AI-assisted application review and candidate fraud detection together transformed the recruiting function — and neither would have delivered full value without the other.
2. Centralize AI-assisted application review configuration to your RecOps team. When criteria are set up before a job reaches recruiters, there's no implementation barrier — recruiters just use the product.
3. Skip company-wide rollout trainings. Instead, do small group roadshows by recruiter archetype (engineering, go-to-market, etc.). Different recruiter types have different questions, and hands-on sessions in small groups dramatically increase adoption.
4. Use the criteria quick filter to surface must-have and must-not-have candidates at volume. This allows recruiters to work through large pipelines in a structured, high-confidence way rather than reviewing every profile sequentially.
5. Vanta learned about candidate fraud the hard way — they hired a fraudulent go-to-market candidate who had no video on, was no-showing meetings, and whose laptop was sent to a different location than their stated address. They have not hired a fraudulent candidate since implementing Ashby fraud detection.
6. Fraud signals — geographic mismatch, low digital footprint, automation signals — travel with the candidate profile throughout the process, so early flags are never lost even if a recruiter initially marks them as not fraud.
7. Vanta's roadmap for agents: replacing workflows currently running outside Ashby (in other AI tools) with in-app agents, including pipeline reporting by recruiter and a closing strategy agent that synthesizes everything in a candidate's profile to help recruiters make the final push.

## **Transcript**

*Note: transcript is auto-generated and may contain minor inaccuracies.*

**Nick Rianoshek**

We're gonna be chatting today about the AI hiring lifecycle. Before I jump into that though, I want to pass it over to Liz. Tell us a little bit about the Vanta team structure and what recruiting looks like there.

**Liz Wagner**

So like Nick mentioned, I lead the ops team at Vanta. We're a Series D security and compliance company. Our recruiting team recruits all sorts of roles — tech, go-to-market, G&A. We're hiring on average 200 to 250 new hires a quarter, and that's been consistent for the last year and a half or so. We've been on Ashby for about three years.

**Nick Rianoshek**

Okay, so hiring a lot — and that's going to really play into what we chat about today. There are four topics we're going to discuss: past, present, future, and takeaways. First, what was Vanta running into? Then, how is Vanta using AI and Ashby today? Third, the future — what are they planning to do? And lastly, you're going to hear from me on things you can do as you leave here. Liz, would love to hear from you about where Vanta was and what you guys were bumping up against.

**Liz Wagner**

We've adopted almost all of Ashby's opt-in AI features. I think our recruiters, before these features, were buried in high-volume applications. I was just looking at our data — these days we get about 10,000 applications a week. So our recruiters were buried in these. Sourcing output was low as a result — I think only 10% of our hires were sourced before we implemented some of these features. And then about this time last year, early 2025, candidate fraud started becoming a real issue. I've heard that from a lot of people. Adopting these features has really changed the recruiting landscape for us. AI isn't just one workflow for us — it's embedded in all stages of our recruiting process. Everything from sourcing, application review, scheduling. A few features that we use day to day: AI-assisted app review, candidate fraud detection, AI feedback summaries, and AI-assisted reporting.

**Nick Rianoshek**

Was there anyone in particular that your recruiters were resistant to at first?

**Liz Wagner**

Our recruiters were a bit resistant to the AI-assisted app review. They were really interested in reviewing every candidate one by one, and defaulting to AI to surface certain candidate profiles just felt a bit unnatural initially. But once they saw the shortlist and kind of put it into practice, adoption slowly grew. They definitely wanted to see the value before the adoption on that piece.

**Nick Rianoshek**

So let's dive into AI-assisted application review. Can you give us a rundown on where Vanta was before you started adopting this and where you're at today?

**Liz Wagner**

Like I said, a lot of time reviewing applications. If we're seeing 10,000 a week, recruiters are spending hours. I think we did two things that were unique. We actually started centralizing the configuration of AI-assisted app review to the ops team. Our RCs on ops configure all jobs for our recruiters. So in the AI features tab, we started going in there and configuring criteria on every single job based on the job description. And then when we sent it back to the recruiters, there wasn't that barrier to setup — there was no implementation. All they needed to do was use the product. And the other thing — which we've taken this approach for a few different product highlights — is what I call a RecOps Roadshow. We have about 60 people on our recruiting team. So we do small group training sessions for them to bring this feature, really tactically, space to ask questions, apply it to their pipeline. I got very different questions about AI-assisted app review from an engineering recruiter versus a go-to-market recruiter. That really helped — the hands-on training to see adoption shift.

**Nick Rianoshek**

So for this one, I'm going to step us through how you get this enabled in your instance. It's very easy — org setup, drop into opt-in features, and you'll see the full breadth of the different AI tools that we have. There are two items we're going to hit on: AI-assisted application review and fraud detection. Easy enough to just toggle these on, and you'll have the ability to turn fraud detection on either instance-wide or job by job for your highest risk roles. How does this actually look when you have a job set up? Liz, would love to hear how you guys are doing this at Vanta.

**Liz Wagner**

So again, we centralize setting this up to the RecOps team. We pop into the AI features tab, paste the job posting in, and press setup criteria. We've actually found Ashby's suggestions quite good. So our RCs don't even try to come up with the criteria — we'll just go through and add all the suggestions. What this is doing is reading the job description, and the AI is automatically coming up with the criteria it'll rate the candidates on. Then if you press validate in the bottom corner — and this is important — it's going to give us a good signal on those criteria, including guideline warnings and EEO warnings. This is going to really help ensure that you're assessing applications against the right criteria.

**Nick Rianoshek**

So when you actually jump into a job, in the application review portion, when we go into bulk review, this is going to pull up the AI-generated criteria on the left-hand side. Not only are we going to see where the candidate met the criteria, where they did not, but also where we're undecided. You can drop down and take a look at the reasoning behind it — it's not just a yes or no, it's really pulling from what you would put into the prompts and looking at it against the application. Liz, would love to hear how you guys are actually using this further.

**Liz Wagner**

One thing we really like is the quick filter at the top. So let's say you've got five criteria but two are absolute must-haves. You can check those two off and it will filter a pipeline of a thousand applicants and just show you those people that meet those two non-negotiable criteria. You can go through those, maybe move them forward really quickly. And then you can actually do the inverse also — find that bucket of candidates that does not meet your two must-have criteria, go through those maybe a bit quicker. We always have our recruiters still click through and validate the AI, but they learn to trust it. This has really helped with just getting through applications on a Friday or a Monday and surfacing the top 10% and bottom 10%.

**Nick Rianoshek**

Let's shift gears and jump into candidate fraud. Great origin story that Vanta has for this one. Would love to hear a bit about it from you.

**Liz Wagner**

Before we implemented candidate fraud detection — I think we got early access last fall — recruiters were constantly asking, is this person real? Poking around, asking their peers to validate a profile. It was all gut instinct, no concrete evidence. We learned this the hard way. We had quite a few examples of candidate fraud in our pipeline. But last spring we actually hired somebody that was a fraudulent candidate. On the go-to-market side — we thought a lot of our fraudulent candidates were on the engineering side, but turns out we hired a go-to-market candidate. They joined, we sent a laptop, and their first week their manager was suspicious. They didn't have video on, they were no-showing meetings and onboarding sessions. He flagged it to the security team, they looked into it and noticed a location mismatch — the laptop was sent to a different place than where we thought the candidate lived. And the team also went back and looked manually to see a low digital footprint — their LinkedIn had just been created two weeks before we hired them. We're a security and compliance company, so obviously this is a big security risk. We sent them a laptop, they got access to all systems. Ashby has enabled us with more confidence. We have not hired a fraudulent candidate since implementing this.

**Nick Rianoshek**

So when we jump into the review in bulk with fraud detection turned on, when a candidate does not throw fraud signals, we're not going to see that on their profile. When a candidate does throw signals, we'll see them called out above the AI-generated criteria — an AI summary at the top so recruiters can have a quick glimpse. We have social signals, automation signals, and location mismatch. For each of these, you can drop down and it gives you the reasoning behind why it's throwing that signal. And the great piece about this is fraud signals travel on the candidate profile. So if we mark as not fraud early on but start getting other flags about them later, we can see the fraud signals have stayed on the candidate profile. You're not losing sight of what might potentially be signals at the beginning even if everything else starts throwing a warning sign.

**Liz Wagner**

Geographic signals always require human review, because it could just be somebody traveling. Like, you could see they're in France and their resume says United States, but it could just be the geographic signal from when they applied. The digital footprint is big — we've found that's a really telling sign if accounts have all been created right before they applied. And it even shows up if they're trying to impersonate another person. Everything else about their profile, the photo, will be really detailed because they will have replicated another LinkedIn profile that is real, but the digital footprint is surfaced here, which has been really helpful.

**Nick Rianoshek**

Let's look beyond what you're doing today at Vanta and really try to understand what you're doing next.

**Liz Wagner**

We heard about AI agents this morning. We've been building a lot of AI agents outside of Ashby. But we're really excited to build them in-app. Because we're using an API now, but everything's in Ashby — our candidate notes, our feedback, our interview scores. So it will make a good foundation for building in-app agents that connect to Slack and do similar workflows agentically and just don't require as much human recruiter intervention to move candidates forward and surface insights. I'm trying to get my recruiters to spend less time navigating the system, more time with candidates, doing the things that require recruiter expertise and human judgment. Two examples I'm excited about: one, an agent that says show me all the candidates a recruiter screened this week or the pipeline for a specific role, and it'll surface that automatically. And the other, which we actually have outside of Ashby and are going to build inside, is the ability for an agent to help craft a closing strategy for a candidate. Candidates at offer stage — the feedback's in there, the recruiter screen notes are in there, everything's in the profile. This agent could help us craft that closing strategy.

**Nick Rianoshek**

There are three things you can take away from today. First, turn on fraud detection — this can be organization-wide or for those high-risk roles where you're already seeing signals. It's a quick toggle and there's a lot of unlock there. Second, configure AI-assisted application review — but don't just leave it to your recruiters to do. Take it upon yourself, your RecOps team, and then roll it out for them. Third, and this is directly from Vanta's approach: skip the large team-wide trainings and really jump in. Go into specific teams and specific groups, put demos in there, understand their exact use cases. Going slow here ultimately accelerates things in the end. Liz, huge thanks for today.

See how Ashby's AI tools work together across sourcing, scheduling, interviewing, and reporting to keep hiring moving. Learn where AI removes manual work and where human judgment still drives the final call.

Ashby

ashby

Nick Rianoshek

Customer Success Manager

AI Across the Hiring Lifecycle

Speakers

Recommended Sessions

What's New in Ashby

Designing for AI Impact in Talent

Product

Resources

Company

Developers