The only measure of AI in recruiting - are you having better conversations?

It is no secret that I hate the very idea of AI interviewers. I am not alone - my candidates almost entirely reject the idea of sitting through an interview with some Jack, Jill, Joe or Joanna or whatever they're now called. If you can't be bothered talking to me (they say), I can't be bothered interviewing with you.

Perhaps there is a place for the AI interview, but it's certainly not in the startup space where earning a candidate's trust is the most important element of the process.

And yet... AI is now the operating system behind many businesses, including ours. We are experiencing one of the fastest accelerations in productivity seen for decades. The only limit is your imagination (and your token budget).

Over the last few months, we have built, and built, and built:

• We built our own CRM, built entirely around the way we work here
• We re-factored our whole database, scoring dynamically
• We developed a sourcing and scouting tool - six agents that search, parse, scrutinise, score & select
• We built automation processes to remove the routine from our day-to-day
• We built an intelligence system to flow meeting notes, market intel, etc. into our day-to-day

The net result is not just that we've eliminated the boring, slow parts of our recruitment processes - it's that we've made them better.

The objective was never "just do this", it was "let's make it better than how we'd have done it". Quicker is a bonus. Get me to a point where I'm able to have better conversations.

So our process is quite simple:

• We ask AI to do the sourcing for us, with as much detail and input as possible (briefing calls, company research, market maps, ideal candidate profiles, etc.) - and it can take hours
• The agents write reports on every candidate, scoring them according to the hard requirements and nice-to-haves, with rich write-ups on everyone, questions to ask
• I view the long-list (often 50-100 candidates) through a variety of lenses - requirement scores, rationale for selection, pedigree, evidence, past notes we've made on them, and I move them along in my process. This is a multiple cup-of-tea session. I develop my own shortlist from this.
• I prepare the outreach sequences, AI prepares the calendar links, bespoke job description links, interview prep notes for candidates... our system manages the sequence
• And then... I have the conversations

Is it quicker? Of course. Is it better? A hundredfold.

The measure is the number of conversations that turn into first interviews with our clients, and that is currently running at 72%. This number includes those that we don't present (the intermediate stage here).

The best conversations with potential candidates are those where:

• this role is a good next step for them
• the approach is well thought out
• the conversation is informed

People talk about "human in the loop", but if anything, it's "AI in the loop" when it comes to recruitment. At our end of the scale, this is about knowledge and relationships - knowledge of the roles because we've done them ourselves, and relationships with candidates, because we're able to talk on a level. Why would we allow a Jack or a Jill to hold a conversation in our place? How insulting would that be to the founding marketer we've been tracking for the last 6 months?

"We're so interested in talking to you that we've sent an AI agent to do it for us" - no thanks.

The discussion about AI in recruitment should never be about the glitzy tools or the fancy functionality, it should always start and end with "am I having better conversations"?