How I Used AI to Beat the ATS and Land Interviews

I’ve applied for hundreds of jobs. Tweaked my CV. Changed the fonts. Added metrics. Cut the fluff. Still - nothing. I was being rejected from jobs I was clearly qualified for. The emails all sounded the same, and they arrived at 2:37am on a Sunday. "Thanks for your time and interest in [Company]. Unfortunately, we won’t be moving forward." No context. No feedback. No human reading it, clearly. So I got curious.

🤖 ATS ≠ AI

It turns out most companies use Applicant Tracking Systems (ATS). Not AI. Not smart. Just keyword matchers - like Google circa 2005. If your CV says Director of Marketing and the job ad says Marketing Director? Rejected. If you upload a PDF with nice formatting? The system reads gibberish and autofills “This section is left blank.” After that, a junior recruiter often screens what’s left - quickly. They're not scanning for nuance. They're looking for keywords. Structure. Familiarity. Not originality. That’s when it clicked: I was talking like a human. I needed to speak robot.

🔍 I needed to understand where this came from. That’s how this infographic was born:

👉 From Paper to Prompts: The Tech That Shaped Recruitment

From Paper to Prompts: The Tech that Shaped the Recruitment Industry

It’s a timeline I made to trace how we got here - from newspaper ads in 1870, to faxes, to Monster, to Gen-AI and automated rejection emails. It's a crash course in why recruitment feels so inhuman now.

Knowing the history helped me realise this isn’t personal. It’s systemic. And solvable.

🛠 So I built a new system

I stopped relying on tools that claimed to “write the perfect CV.” Most of them just regurgitate the job spec and list skills I don’t have. Here’s what actually helped:

• Jobscan – to score my CVs against job descriptions

• ChatGPT – to write keyword-aligned summaries and bullet points

• NotebookLM – to build a memory bank of my experience, old CVs, transcripts, and project examples

• Glean (Genio) – to record interviews (where appropriate and allowed) and meetings so I could reflect and reuse real phrases

Manual testing – A/B testing job title phrasing, formatting, and even location filters to get past the bots

✅ What changed?

I stopped guessing and started experimenting. I learned how ATS systems interpret formatting. How to phrase things like a job spec. How to not say “prefer not to say” on diversity questions (that’s apparently a red flag for some filters).

It’s not about sounding impressive. It’s about sounding matchable.

And once I started writing for the system and the reader, interviews started coming in again.

🪜 Try this prompt:

Paste this into ChatGPT (along with your CV, LinkedIn, and job description):

Create an ATS optimised CV for this role [insert whole job description]: Ensure the subtitle, professional summary, job titles and job descriptions are keyword optimised and match the phrasing in the job spec (e.g if job spec says Marketing Director, use this and NOT Director of Marketing). I am aiming to achieve a JobScan match score of >80%. Don’t be boring and generic, don’t include any descriptions that are mostly fluff, waffle, bullshit or cliches. I want this CV to be written with the mindset that I am the winning candidate for this role, make this CV stand out among the 1500 they are likely to receive, be confident but don’t be cocky or arrogant. Write directly, concisely, and naturally. Avoid extra adverbs such as ‘additionally’, complex phrases, or overly verbose or professional language. It should be reflective, not just observations, focus on how I did it, not just what I did. The summary paragraph should be no more than 100 words or 5 lines of text, the job description needs 3 bullet points per job with a summary of responsibilities, a key achievement and somethingI learned, each bullet point should not exceed one line of text.

It’s not magic. But it’s a massive head start.

🎤 Your turn

I’d love to know what you’ve found.

What’s worked for you? What hasn’t?

Have you gamed an ATS? Or lost a role to a formatting glitch?

Let’s compare notes - and fix this system from the inside.