Where AI fits in the recruitment funnel (and where it doesn't)

Application volumes are rising, CVs are increasingly AI-polished, and the list of tools promising to solve both keeps getting longer. For most TA teams, the challenge isn't knowing that AI is part of the answer. It's knowing which part of your process needs attention, and whether an AI tool is even the right answer.

In episode 8 of the Pinpoint How-To Series, I sat down with Bradley Clark, Head of Talent at Article, and Sean Kennedy, founder of Kennedy's Content, to build a practical framework for thinking through exactly that.

Start with your process, not a tool

Before you evaluate a single AI product, you need to know what problem you're actually trying to solve. Most teams skip this step, and it's the most expensive mistake you can make.

The practical way to do it: map your hiring process end to end and find your biggest bottleneck. Look at where your team is spending the most time. Where are things slowing down? Where are candidates dropping off? Where are hiring managers frustrated? Your ATS data can tell you a lot of this: time to hire by stage, application-to-interview conversion rates, how long roles are sitting open.

Once you've found the bottleneck, ask three questions before you look at any tool:

• Is this a process problem? Something broken in how you work that needs fixing before any tool will help.
• Is this a capacity problem? Your team is stretched, and more tooling without more clarity will make it worse.
• Is this a tool problem? A genuine gap where the right automation or AI would make a real difference.

A lot of the time, the answer to the first question is yes. AI applied to a broken process doesn't fix it. It makes the problems faster and more expensive. Get the foundation right first.

Know the difference between automation and AI

These two things are often talked about as if they're the same. They're not, and reaching for AI when automation is the right answer wastes time and money.

Automation is rule-based. You define the conditions; it executes them reliably every time. Screening applicants against knockout criteria, sending rejection emails with specific reasons, triggering status updates, scheduling interviews: all automation. Predictable, auditable, cheap to run at scale.

AI is judgment-based. It helps you make sense of information, surface context, and work through problems that don't have a simple yes or no answer. Prioritizing a large applicant pool, drafting job descriptions, preparing interview questions based on a candidate's background: this is where AI adds value.

Most teams dealing with high application volume instinctively reach for AI. In most cases, automation is the right first move. Clear the noise, cut the manual work, and give your team time back. Then use AI to work through what remains.

Learn how to set up automations in Pinpoint

A framework for every AI decision you make

Once you've identified a genuine problem that a tool might solve, you need a way to decide what kind of solution fits. One framework that came out of this conversation is worth applying to every AI decision you make: three buckets, drawn from Ryan Roslansky and Anish Laman's book Open to Work.

1. Replace. What can AI handle entirely, without impacting candidate experience or decision quality? Initial sorting of a large applicant pool, drafting routine communications, pulling together background research. AI does the work and a human reviews the output before anything moves forward.
2. Collaborate. What can you do better with AI alongside you? Job descriptions are the obvious starting point: AI can check for bias, improve clarity, and tailor language for specific candidate groups. Briefing hiring managers is another: AI helps you represent candidates more clearly, especially in technical roles where the recruiter and the team don't always share a vocabulary.
3. Keep human. What stays entirely with a person, no exceptions? Final interviews, offer conversations, the moments where a candidate meets the team. Any decision that materially affects someone's livelihood. These are the parts of hiring where judgment and real human interaction are genuinely hard to replace.

Run this test on any tool you're evaluating. If the vendor can't clearly tell you which bucket their product sits in, that's worth knowing.

Where to hold back

A few areas where the honest answer is: not yet, or the risk isn't worth it.

AI screeners and AI interviewers. These are heavily pitched right now, and the technology is developing fast. But putting an unvetted AI in front of candidates creates real risk to candidate experience, your employer brand, and potentially to compliance. Most teams and tools aren't ready for this yet.

Candidate data through external tools. Be careful about what you feed into tools. The questions to ask any vendor: What models do you use? How are they trained? Where does the data live? Are you GDPR compliant? How do you audit usage? A lot of products on the market right now are wrappers on a large language model without clear answers on any of those questions. For anything that touches real candidate data, default to the tools built into your ATS until you've done proper due diligence.

Transparency with candidates. Whatever you're using, say so. Candidates are anxious about AI in hiring, and most of that anxiety is based on assumptions rather than facts. A clear, plain-language statement about what's automated, what's AI-assisted, and what stays human builds trust and makes your process feel fairer.

Key takeaways

• Find your bottleneck before you evaluate any tool. The problem should drive the solution, not the other way around.
• Automation and AI are not the same thing. Use automation to clear the noise first, then use AI to work through what's left.
• Apply the three-bucket test to every tool you evaluate: replace, collaborate, or keep human.
• Be cautious with candidate data. Ask hard questions of vendors and default to tools you already trust until you have clear answers.

What to do this week

1. Run a process audit. Map your hiring process and find where time is actually going. Your ATS data is a good starting point: time to hire by stage, drop-off rates, and how long roles sit open will tell you where to look.
2. Check your automations. If you're not using automated rejections with specific reasons, that's one of the highest-value things you can set up in Pinpoint right now.
3. Apply the three-bucket test to any AI tool you're currently evaluating or being pitched.
4. Draft a candidate-facing statement about how you use AI in your process. Even a short paragraph builds trust.