The market for AI recruitment tools has expanded rapidly over the past couple of years, which is good news in one sense and confusing in another. There are now a lot of products to choose from, and the marketing tends to blur together. Everything claims to be intelligent, fast, and accurate. Knowing which claims to actually test and which questions to ask before you sign up can save you a lot of wasted time and money.
Here is a practical guide to evaluating AI CV screening software, written for UK recruitment agencies who want real results rather than impressive demos.
Does It Understand Context, or Just Keywords?
This is the most important question to ask, and it is one that many tools fail on in practice even if their marketing suggests otherwise.
Keyword filtering has existed in ATS systems for years. It works by checking whether certain words appear in a CV. The problem is that two candidates with identical experience might describe it in completely different ways, and a keyword filter will rank one highly and miss the other entirely.
Genuine AI screening understands what a candidate has done, not just which specific phrases they used to describe it. A good way to test this is to take a handful of CVs where you already know the answer, anonymise them, and run them through the tool. Does the output match your professional judgement? If the best candidates are coming out on top regardless of how they have worded their experience, the AI is working properly.
How Quickly Does It Actually Produce Results?
Speed is one of the headline benefits of AI screening, but there is a meaningful difference between "faster than doing it manually" and "fast enough to change how you work."
If a tool takes 10 minutes to process a batch of 100 CVs, that is faster than a human, but it is not transforming your workflow. If it takes seconds, that genuinely changes what is possible. You can screen and shortlist in the time it used to take just to open the first few CVs.
Ask about processing time before you buy, and test it with your actual volumes rather than the vendor's demonstration examples.
Is It Built for Recruitment Agencies or for In-House HR?
This distinction matters more than it might seem. In-house HR teams and recruitment agencies have different needs. In-house HR is typically screening for one employer's culture and requirements, often for roles they have filled before. Agencies are screening for multiple clients, multiple sectors, and roles that change constantly.
Software built for in-house HR often assumes a fixed set of job families, a standard company culture profile, and a relatively stable set of requirements. That does not translate well to agency life, where you might be screening energy sector engineers on Monday and marketing managers on Friday.
Look for a tool that lets you define the criteria fresh for each role, handles variety across sectors, and does not require a long setup or configuration process before you can run your first screen.
What Does the Output Actually Look Like?
A ranked list of candidates is the basic output, but the best tools give you more than that. You want to know not just who ranked highly, but why. What were the specific strengths that put this candidate at the top? What was missing from the candidates who ranked lower?
That information serves two purposes. First, it helps you sense-check the AI's reasoning and apply your own professional judgement. Second, the key strengths highlighted for top candidates become the foundation of your client submission and coversheet, saving you significant time at that stage of the process too.
If the output is just a score with no explanation, you are getting less than half the value.
How Straightforward Is It to Get Started?
Some recruitment technology requires weeks of onboarding, training, and configuration before you can use it properly. For smaller agencies in particular, that is a significant barrier. If it takes a month to get up and running, most teams will either not bother or will never get past the setup stage.
The better tools in this space are designed to work from day one. You should be able to upload a job description, upload your CVs, and get a useful result without any technical knowledge or lengthy setup. If a demo requires a specialist to guide you through it, that is a signal about how complicated the day-to-day use will be.
What Does It Actually Cost?
Pricing in this space varies significantly. Some platforms charge per CV screened, which can add up quickly on volume roles. Others charge per seat or per month, which is easier to budget for. Enterprise platforms often require an annual contract and a significant upfront commitment.
For most UK agencies, a monthly subscription with no long-term commitment is the sensible starting point. It lets you properly evaluate the tool over a real working period, across real roles, without being locked in.
What Lucuma Offers
Lucuma is designed specifically for UK recruitment agencies and built around these principles. It produces ranked shortlists in seconds, explains the strengths of each top candidate so you can go straight into writing your submission, and can be up and running on your first role within minutes of signing up.
Pricing starts at £99 per month for solo consultants, with agency and enterprise tiers available. There is a 14-day free trial with no card required, which gives you enough time to test it across a few real roles and see the difference for yourself.