I have had this conversation more times in the last two months than any other. Agency leaders calling in, telling me their teams are all over the place with AI. Everyone is using something different. No one is sure if it is actually working. And underneath all of it is a quiet fear that something could go very wrong.
They are not wrong to be worried. The risks are real. But they are also manageable, if you know where to look.
This edition covers the biggest risks I am seeing right now across the agencies we work with, and what to actually do about each one.
Risk one
This is the most common thing I see. An agency starts getting bombarded with AI platforms. Someone tries a new candidate matching tool. Someone else adds another sourcing tool. The leadership team signs up for three different subscriptions in a quarter because every vendor promises it will transform the business.
What actually happens? Consultants get confused. They are spending more time inputting data into these platforms than the tools are saving them. And the tools are not talking to each other, so nothing is syncing and everyone is copying and pasting between systems all day.
"I am yet to see an agency that has implemented AI seamlessly across their processes. The vast majority are in that testing phase of trialing different products. And along with that comes a lot of spend."
Ez Khan, Hume ScopeMy advice is always the same: go slower. Pick one tool. Get one person using it properly. Get actual use cases. Then roll it out.
And before you add anything new, ask yourself one question: does this sync with your existing tech stack? If it does not, and a consultant has to manually move information between platforms, it is almost certainly not worth it. The time it takes to do that defeats the purpose.
The test I use: if we were actually paying for this and not getting it for free, would I still use it? For most of the platforms we trial, the honest answer is no.
Risk two
This one keeps me up at night. I say it in almost every conversation I have with an agency leader and the reaction is always the same: genuine concern, usually followed by "I had no idea."
Here is what is happening right now in most agencies. Consultants are using their personal Claude or ChatGPT accounts to do their work. They are feeding in client meeting transcripts, candidate information, job briefs, sensitive correspondence. All of it is going into a personal plan.
Two problems with this.
If a consultant leaves and they have been using a personal AI account, they take all of that context with them. Every client conversation they loaded, every candidate file they processed. You have zero control over it because it is their account.
This is in the terms and conditions of most personal AI plans, including paid ones. When your consultant loads a client transcript into a personal GPT account, that platform is using your data to train itself. In theory, a competitor asking GPT about your client could surface information that came from your team.
The fix is straightforward. Move the team to a business account. On a business plan, the platform does not train on your data. If a consultant leaves, you can switch off their access immediately. You have control.
For Claude specifically, the business plan starts at three seats and is very cost effective. In most cases the pricing is comparable to what your team is already spending on individual pro plans.
Risk three
The third risk is subtler but it is just as damaging. Most agencies have consultants using AI in their own way, with their own prompts, getting inconsistent results. One consultant gets great candidate profiles. Another gets something generic that needs to be rewritten from scratch. The agency cannot scale anything because nothing is standardised.
The agencies that are actually getting return from AI have done one thing differently. They mapped their existing process first, identified exactly which steps AI could do faster and better, and then built those steps into a repeatable skill or workflow. The consultant does not have to think about what to prompt. They just run the skill.
We helped one agency set this up for their job briefing process. The consultant picks up a new job, runs a single skill, and out comes a job description, a re-brief email to the client, three candidate profiles, social posts, approach emails, and candidate outreach. All done. All in their format. One command.
That is not magic. It is process work done upfront so the AI can do it consistently every time.
What to do this week
Do an audit of what your team is actually using.
Which platforms. Which accounts. Who is paying for them. Personal or business plans. You cannot manage what you cannot see, and most leaders are genuinely shocked by what they find when they ask.
Move to a business account before you do anything else.
Get the team off personal plans. For Claude, three seats minimum on the business plan. This is not optional if you care about your data. Do this before you invest any more time in training or workflows.
Stop trialing new tools until you have one working properly.
Pick the platform that integrates best with your existing tech stack. Get one or two senior people genuinely using it well. Build the use cases. Then think about rolling it out. Adding more tools before you have nailed the first one is how you end up with confused consultants and zero ROI.
Map one process before you build anything.
Pick the highest volume, most repetitive part of your recruitment process. Write down every step. Then work out which of those steps AI can do faster and better than your consultants. That is where you build your first workflow. Start there and prove the value before you expand.
How we can help
The most common thing I hear from agency leaders is that they know they should be doing more with AI, they just do not know where to start or whether they are doing it right. That is exactly what we built our two offerings around.
This is the fastest way to get your team set up properly, without wasting months figuring it out yourself. 16 sessions, each under 10 minutes, with prompts and downloads to implement as you go.
Right now it is $249/mo AUD for your first seat. On 1 September that goes to $498. No lock-in, cancel any time, but the rate you start on is the rate you keep.
For agencies that want hands-on support. We come in, audit your current setup, map your workflows, make recommendations, and help you implement. Usually starts with a three-hour session.
Final thought
AI in recruitment is not going anywhere. Every agency leader I speak to knows that. But right now, the agencies that are winning are not the ones using the most tools. They are the ones that picked one, got it working properly, and built actual processes around it.
The wild west phase will not last forever. The agencies that get their foundations right now are the ones that will pull ahead when everyone else is still trialing their fourth platform and wondering why nothing is sticking.
Book a free 30-minute discovery call. We will walk through where things are breaking down and what to fix first.
Always happy to have a chat,
Ez Khan
