The Rise of AI Screening Tools And Why We Still Believe in Human Judgement

Every few months, a new tool lands in our inbox promising to change how we hire. AI CV screening. Automated shortlisting. Algorithms that claim to predict, from a CV alone, who will succeed in a Dynamics 365 role and who won’t.

Some of these tools are useful. Most of them are solving the wrong problem, or causing new ones.

What AI screening actually does well

Let’s give credit where it’s due. AI tools are fast at pattern matching. Feed them a job description and a stack of CVs, and they’ll flag keyword overlaps, certifications, years of experience, and tech stack matches in seconds. For high-volume, low-differentiation roles, that can be a real time saver.

We use technology in our own process too because nobody’s manually cross-referencing candidate databases by hand in 2026. Some tools have their place.

Where it falls apart in D365 recruitment

Here’s the problem specific to our corner of the market: Dynamics 365 roles are rarely about keyword matching. Two candidates can have near-identical CVs: same certifications, same years of experience, same module specialisms, but be completely different hires.

One might have spent three years on stable, well-scoped implementations. The other might have spent three years firefighting on projects that were in trouble from day one. Same CV. Wildly different candidates. An algorithm scanning for “D365 F&O, 3 years, Power Platform” sees no difference. A recruiter who’s spoken to both of them does.

AI also struggles with the things that actually predict success in consulting-heavy roles: how someone communicates with a difficult client, whether they can hold their ground in a scoping conversation, how they behave when a go-live date slips. None of that shows up on a CV. All of it shows up in a fifteen-minute phone call.

The bias problem nobody likes to talk about

AI screening tools are trained on historical hiring data. If that data reflects biased hiring patterns, and in tech, it often does, the tool doesn’t correct the bias. It automates it, at scale, while wearing a coat of algorithmic neutrality that makes it harder to challenge than a biased human ever was.

We’ve seen candidates get filtered out by keyword-matching tools simply because they described their experience differently than the job spec was phrased, not because they lacked the experience. A gap in employment for a legitimate reason. A career change that doesn’t fit a template. A non-linear path into D365 from a different ERP background. These are exactly the candidates a good recruiter digs into, and exactly the candidates an algorithm quietly discards.

What we actually do instead

Every CV that comes through Shape IT gets read by a person who understands what a Dynamics 365 project actually looks like from the inside. We call candidates. We ask about the projects that didn’t go well, not just the ones that did. We ask clients what “senior consultant” really means to them for this specific role, because it’s never the same answer twice.

That’s slower than running a CV through a filter. It’s also the reason our placements tend to stick.

Where this is heading

We’re not anti-technology, and we’re not going to pretend the tools won’t keep improving. But for roles where judgement, communication, and project context matter as much as technical skill, we think the recruiter’s job is to add the layer that AI can’t, not to hand the decision over to it.

If you’re hiring for a D365 role, or looking for your next one, you’ll always talk to a human at Shape IT. We think that’s still the point. Let’s talk.