Everyone 'wants something to do with AI', but too often implementation is an end in itself. Guillaume Onclin has been active in recruitment for eight years and is now working as an independent consultant. He sees daily where the profit lies: not in separate tools, but in processes that are AI-native. "Implementing it properly so that it actually adds value is still a lot more difficult. But a world without AI is already hard to imagine these days."
In this article you can read his vision, the three maturity phases he sees in organisations, the real benefits (and risks), and how to go from experiment to scalable impact.
Recognise the three phases and consciously choose what you are working towards
Guillaume recognises three distinct phases in how organisations embrace AI:
- Phase 1, experimentation by recruiters: recruiters try out free tools to rewrite texts or summarise CVs, often without a framework or policy. According to Guillaume, "recruiters are mostly experimenting on their own initiative." The result is fragmentation and risk of data leaks.
- Phase 2, the company procures tools: Official AI solutions emerge, only the work remains manual. There is still copying and pasting between systems, making AI an extra step rather than an accelerator. As Guillaume explains, "Companies offer AI tools that you can use, but a lot of the work still lies with the recruiter."
- Phase 3, AI is part of the workflow: AI is linked to your ATS and automations. When creating a job posting, draft texts, target group variants and an initial shortlist automatically appear. He explains: "As soon as you add a role to your ATS, AI-driven steps happen automatically." This is the target image, this is where the scalability is.
Why the leap from 2 to 3 is stalling (and how to make it anyway)
The will is there, the integration is not. Many vendors build functions for the individual user, not for links with APIs and webhooks. As a result, AI remains something you operate alongside. Guillaume notes, "Vendors focus on immediately deployable functions for recruiters, not on how those functions become part of automations. The breakthrough comes when you link AI to process events, e.g. vacancy created, interview scheduled, placement completed, and the system automatically triggers the right tasks at that point."
The real payoff: time back, better data, stronger conversations
The most tangible gain is time. He stresses, "AI brings a gigantic efficiency gain." In addition, data quality increases because you capture more and more consistently. And conversations get better, before and after the interview. In his words, "AI helps with preparation and evaluation, what questions did you miss, what signals did the candidate give, what follow-up makes sense." At the same time, he warns of privacy risks if you keep experimenting with free tools. Guillaume says: "Recruiters use AI anyway, you'd rather use it in-house because of the privacy of clients and candidates."
Think candidate first, AI as a service for the applicant
The conversation about AI is often about productivity for the recruiter. Guillaume wants to broaden the perspective. He states, "I would love it if we talked more about how to make the experience of customers and candidates better. A concrete example is an AI-driven practice environment for candidates, sent automatically after the interview confirmation and tailored to the role. Candidates practice questions, hone answers and come in better prepared. That is noticeable value, not just faster, but more importantly better."
Roadmap from phase 1 to phase 3, without detours
- Make AI secure and central: Choose enterprise solutions and ban free tools for sensitive data. He stresses, "You'd rather use it in-house because of privacy."
- Invest in data quality: standardise fields, enforce data completeness, record decisions and notes uniformly. "That need has only increased," he says.
- Automate the undercurrent: Link AI to your key process events. Start with two or three critical flows, such as job publication and interview planning, and then scale up.
- Build deeper, not wider: Fewer tools, more integration. He warns: "Avoid a fragmented landscape with different tools, make AI a real part of automations."
- Measure what matters: Track time to shortlist, data completeness, candidate NPS, quality of hire signals and interview preparation time.
- Design for the candidate: Standardise updates, preparation and feedback. See AI as a service, not a gimmick.
Technology is means, process is king
New technology is never the goal in itself. Guillaume sums it up: "It's about how you want to work, you look for the right technology to go with that, the technology is a tool." So start with the flow you want to improve, for recruiter, hiring manager and candidate. Then choose the tooling that makes that possible and link AI to your process events. Start with one job route from start to finish and automate it, if it works well, roll it out. This is how you move from buzz to business and reach phase 3, AI as the engine of your recruitment process.
