Our Chief Operations Officer, Harry Elkington, reflects on the pitfalls and overall impact of AI on recruitment.

Over the last six months, I have had countless discussions with AI and automation providers to see how we can plug in their technology to enhance our current process management and talent origination.

There is no shortage of companies promising efficiency gains or revenue uplift, AI is hot right now and everyone in the recruitment industry seems to be fighting to implement a new product. However, a lot of these AI/ Automation businesses, if I’m honest, are playing in broadly the same ballpark.  There is a significant cross over between the main automation platforms and recruitment businesses should be cautious as to the length of contract they are committing to, especially with the pace the industry is moving at and all these top providers fighting for market share. But the position I keep coming back to is, as exciting and seemingly game changing as all of these plugins may be, they are useless without structured and high integrity data.

Looking after our precious data

If a recruitment firm isn’t obsessive about internal coding, system processes and how it captures its own IP, then a lot of the upside from these tools just won’t land. You can’t bolt automation onto a rubbish pile and expect magic.

Coding candidates by skillset such as ACA, PE-backed, SaaS, or Series B is useful if it helps consultants find people quickly when a brief comes in. But this technology is still fairly surface level. While it can tell you what someone is qualified as, it does not necessarily reflect what they’ve actually delivered. This can be a critical factor in determining if a candidate is right, or not, for a specific role.

Where this gets more interesting, for us at least, is how we use our ‘conversation data’.

All of our calls are now recorded and categorised by our AI assistant, which means that all human-to-human interactions are converted into a human-to-machine format. This enables our teams to search way beyond skill codes alone.

For instance, if a client asks for someone who has led a turnaround of a failing division, or delivered £12 million revenue growth post-acquisition, we can search across every structured conversation we have had with a candidate in the last 12 months. That’s using our IP properly. It’s a different level of retrieval than just filtering by keywords on a CV.

AI as an enabler 

AI may have its limitations in the here and now, but as the technology evolves so will the opportunities for businesses like ours, who will have that all-important competitive edge because we’ve embraced AI’s capabilities and evolved our operations in sync with the technology.

The most transactional, lower skilled roles will obviously be absorbed first. That’s already happening. I know engineering recruiters in the US construction recruitment space clearing more blue-collar workers in an hour than they could in a day just a year ago. That’s great for them right now, but my concern is there is nothing stopping the end client doing the same thing internally once they invest in the software and understand how to use it. That gap is closing quickly.  I think the five-year topic on ‘AI in recruitment’ isn’t really AI versus recruiters; it’s better recruiters using AI versus recruiters who don’t evolve.

So, the real question is, what are we selling that an internal recruitment function with AI cannot replicate?

If it’s just outreach, screening and moving CVs around faster, then fees will come under pressure. As sourcing becomes easier and cheaper, clients will expect more. Margin compression is bound to happen in parts of the market.

However, if it’s judgement, real market knowledge, understanding competitor movements, advising on compensation and market comparison, knowing who’s actually credible and who isn’t, knowing which teams have terrible culture or drive people relentlessly 16 hours a day, that’s different. That’s harder to replicate internally. I don’t think it’s impossible but without a structured amount of compounded data over a good frame of time it won’t be reliable.

Who speaks in keywords?

At senior level especially, experience matters. Many Private Equity executives we’ve worked with have gone through AI screening and automated interviews and hated it. The higher up you go, the more essential meaningful human relationships become. Leaders want to engage with real people who understand their motivations and aspirations – not feel processed by a system. AI increases speed. It doesn’t automatically increase trust.

Another thing I keep thinking about is whether we’re just renting capability. What I mean by that is if every recruitment firm buys the same plug-ins, then nobody has an advantage. The difference comes from how good your underlying data is and how well it’s structured. The tool isn’t the edge – the dataset and IP are.

Over the last six months we’ve implemented a lot of this internally, tightening coding, structuring notes properly, making sure we’re capturing more than just surface information. The next stage for H&P is thinking longer term. What data will we wish we’d been capturing in five years? What patterns of success are we not tagging yet? How do we turn daily conversations into something that compounds over time?

If AI handles the mechanics, what’s left is market insight, context and judgement. That’s what we need to keep refining. That being said, if and (dare I say inevitably) when, an AI assistant knows more about the market, can operate instantly, for a fraction of the cost, and in the voice of Homer Simpson (other voices I suspect will be available), I think we all have to admit, we’re going have to rethink the value proposition of what recruiters can bring to the table.