The role of a recruiter has become so broad in recent years and the pressure is so high that new technology is welcomed with open arms. However, nothing is as frustrating as when you think you’re boosting efficiency with AI, only to find yourself having to make adjustments, corrections and work across multiple systems more often than you’d like. Do you want a well-functioning AI colleague? It all starts with training and context.
TL;DR | Many AI tools in recruitment use a generic model that isn’t tailored to your organisation. This leads to chatbots striking the wrong tone, giving incorrect answers and confusing candidates. The solution lies not in better AI, but in better context: information about your processes, culture, tone of voice and selection criteria. The more you provide an AI colleague with, the more tasks you can entrust to it. When choosing an AI tool, therefore, look for the ability to add your own knowledge base, customise instructions and integrate with your ATS.
One size does not fit all – why generic AI models don’t work
AI is smart, but only if you make it smart by providing the right context. With a generic AI model – one that hasn’t been trained or adapted to your processes and culture – you’ll quickly miss the mark with candidates. Generic (AI) chatbots use the wrong tone in their messages, give incorrect answers to questions, or make their responses so generic that they don’t actually help the candidate. The result: the candidate gives up.
This isn’t because AI doesn’t work for recruitment, but because AI without background information is, by definition, generic. If you tell the AI tool how your processes work, what the culture is within your organisation, what the right tone of voice is, which target groups you are approaching, what the key selection criteria are and what the roles entail exactly, you will have very different conversations with candidates.
“We used to work with a static chatbot, but that really only caused problems. It caused confusion among candidates, meaning the helpdesk spent a lot of time dealing with all the queries that came in. We even ended up switching the chatbot off. With our MrWork AI Agent, it’s completely different. I’m amazed at how smart our new digital colleague is and how quickly Abbi learns.” – Patrick Friesen, Manager of Marketing, Communications and Partnerships at AB Werkt
AI agents you can build and train yourself
That context – or the knowledge base on which the AI relies – makes the difference between a generic AI chatbot and an AI colleague who acts as an extension of the recruitment team. You therefore want to use AI tools where you can add that context yourself and train the AI colleague, for example by monitoring conversations and then adjusting the prompts (instructions). This ensures that candidates who ask a question about the application process are well informed and that you can rely on the pre-screening carried out by an AI colleague on your behalf. In short: a better candidate experience and less manual work for you.
“It’s a relief that our AI colleague Vivian is always there for our candidates. We receive lots of responses every day, and Vivian handles them in a friendly and helpful manner. We monitor those conversations and can easily fine-tune and optimise them.” – Amy Narold, Recruitment Marketer at ViVa! Care Group
The more context you provide your AI colleague, the more tasks you can entrust to them. From finding the right vacancy for a candidate to pre-screening and even helping to prepare for a job interview.
Points to consider when implementing Agentic AI for recruitment
When choosing and implementing an AI recruitment tool, you should therefore pay attention to:
- The ability to add an extensive knowledge base that provides the context for what the AI agents say and do
- The option to customise the AI agent’s instructions or tasks so that they align with your processes and working methods
- Integration with the ATS – otherwise you’ll keep switching between multiple systems and wasting time
“Most AI tools run on a single generic model that can talk, but doesn’t understand the organisation. Our AI Hub is built around context and control: company-specific knowledge, rules of conduct and processes that are technically embedded in the architecture. This results not in a generic chatbot, but in an AI agent that works predictably and reliably within your organisation.” – Zeno Lampe, CTO at MrWork
What you put in determines what comes out
With AI, it’s a case of input determines output. So when choosing an AI tool, make sure you look not only at what it can do, but also at what you can feed into it. An extensive knowledge base, the ability to customise instructions and good ATS integration are not just nice-to-haves. They are the prerequisites for enabling AI to function as an extension of your team.
