Expectations for AI in recruitment are high for 2026, as this year marks the shift from trial and error to systematic implementation. At the start of this year, 90 per cent of recruitment professionals said they were using AI, and 81 per cent planned to invest more in this smart technology than last year. Impressive figures! Halfway through the year, it’s time for a mid-year review: do expectations match reality, and what lessons can we learn from the first half of this year? In this article, we look back and ahead, together with Mees van Velzen.
TLDR | AI in recruitment has moved beyond the experimental phase: 90 per cent of employers are already actively using it and 81 per cent are investing extra in AI this year. With time savings of 20% and recruitment costs reduced by up to 40%, research by LinkedIn, the Intelligence Group and DataRefs confirms the impact of AI. As well as greater efficiency, a better candidate experience and a higher application-to-hire ratio, there are also lessons to be learnt from the first half of 2026. For instance, context is essential for an AI agent to function effectively; you need to set aside time for training AI colleagues; and it is important to consider security, integration options and training opportunities when choosing an AI tool. New AI software is being developed at breakneck speed, but according to Gartner, almost half of it will have disappeared from the market within 1.5 years.
High expectations
If one thing was abundantly clear from the MrWork AI Trends Report 2026, it was that AI is now truly making its mark in recruitment. The experimental phase is over; many employers and agencies are already using AI and are continuing to expand its use.
- 90 per cent of employers use AI in their recruitment processes. So far, this has mainly involved searching for candidates, making contact and drafting job adverts.
- 78 per cent of employers who use AI in their recruitment process say that this has reduced their time-to-fill.
- 25 per cent of recruitment professionals cite optimising internal processes using technology as their top priority. This makes it the most important item on recruiters’ agendas, ahead of investing in employer branding or improving the candidate experience. Two priorities that, until this year, had always topped the list.
The biggest challenge still lies in striking a balance between the quality and quantity of candidates. This was highlighted in the AI Trends Report 2026 and was once again evident from the poll conducted during the MrWork webinar ‘Ahead or behind? The AI check for recruitment teams’ in July, where participants ranked this challenge as their number one priority.
The current situation: AI is gaining ground
We’re already halfway through the year, and various studies show that expectations largely match everyday reality.
Research by LinkedIn shows that teams using AI achieve an average time saving of 20 per cent. At Intelligence Group, they’ve found that 91% of recruitment professionals use AI on a daily basis, and DataRefs has measured a clear impact from this: a 30% to 40% reduction in recruitment costs and a 50% reduction in time-to-hire.
The most commonly used AI tools are still the chatbots based on large language models (LLMs), such as ChatGPT, Claude, Gemini, and so on. The AI hiring assistant within LinkedIn Recruiter is also frequently mentioned, as are AI features built into the ATS. This latter category primarily offers solutions in the areas of searching and matching.
Some notable high-growth companies in the AI recruitment software market:
- In2Dialog and Carv for recording and analysing interviews
- MrWork for deploying AI recruitment colleagues
- Robin, SeekOut, Fetcher and Moonhub for sourcing
Candidates aren’t standing still either. At the start of 2026, it was still limited, but automated job applications via AI tools such as Laboro, Jobcopilot and SimpleApply are on the rise. In the webinar, Mees van Velzen, founder and CEO of MrWork, explains: “As more and more candidates start applying automatically, the volume of responses will increase significantly. And so the question is: how do you, as a recruiter, best deal with this? Thorough screening – for both soft and hard skills – becomes even more important. We are monitoring these developments closely and taking them into account as we continue to develop our software.”
The results speak for themselves, but do remain critical
MrWork colleagues are now working at 100 different organisations and have already conducted hundreds of thousands of interviews. This has yielded important insights and lessons learnt.
- Responding to candidates straight away significantly improves the candidate experience. It’s also much more pleasant for the recruiter not to have to wade through dozens of replies and questions first thing on a Monday morning.
- On average, AI colleagues save each recruiter 8 hours a week. Sometimes the savings are even greater. FloraHolland, for example, has already followed up with 7,000 candidates via its AI agent – responses that would otherwise have had to be dealt with manually.
- A larger proportion of applicants are actually being hired. This improvement in the application-to-hire ratio is likely due to two factors: candidates’ interest is being better retained, resulting in fewer dropouts during the process, and recruiters have more scope to speed up the human aspect of the process.
All this experience and data suggests that AI without context is, in fact, already a thing of the past. “The promise of AI in recruitment isn’t just about saving time. It’s also about delivering a better experience for both candidates and recruiters. To fulfil that promise, your AI agents need context. Only then can they function effectively and contribute to a better candidate and recruiter experience,” says Mees.
Another point to bear in mind: AI agents are just like new colleagues. Even though you can get them up and running quickly, their training is key to their success. If you’re going to implement AI recruitment colleagues, set aside time for monitoring and fine-tuning. That way, you’ll get the most out of them.
Mees concludes the webinar with a prediction from research and consultancy firm Gartner: “With new technologies advancing at such a rapid pace, you’ll see new tools popping up in no time at all. The same applies to AI. Thanks to vibe coding, you can build these kinds of applications very quickly, and a demo might look promising, but the problem often lies in security or integration with your other systems. Gartner therefore predicts that 40 per cent of AI projects currently being launched will have been scrapped by the end of 2027 due to rising costs, insufficient added value or poor security. You should therefore take a critical look at providers of AI tools, as some of these players will have disappeared from the market within two years.”
This means that procuring AI software is actually becoming both easier and more difficult at the same time. The range of options is growing, but well-integrated tools that link to your ATS and adapt to your processes will remain scarce for at least another two years. Those who are already making careful selections are leading the way – both now and in two years’ time.
