Note: Trying to get this week’s header image to be fun and positive, but not ick? Not easy, and not sure I managed it TBH. At least there weren’t love hearts in this version.

Pretty much all the coverage we see around AI in HR is on things like job applicant trackers, and CVs being screened by AI and good people not making it through due to clueless keyword checks (with a lot of recruiter denial). There’s obviously a lot more to it than that, but even then, the coverage tends to skew towards dedicated apps, and large corporate use.

There are ways, though, to use the standard (free version) LLMs for HR and job-type tasks, just (as always) making sure you’re practicing safe data.

Job Applications – Applicant Side

First, looking at the non-company side of the equation. If you’re looking for a job, an LLM can be a huge help in getting your ducks in a row as you work out what you want, what you’re suited to, and how to present your case.

Start by pulling your contact info (and any other Personally Identifying Information) off your CV, putting it into one of the LLMs and asking something like: Looking at the attached CV, can you please suggest the types of roles and businesses I should be focused on in any future job search?

I already know I’m better suited to larger corporations, because that’s where I’ve spent much of my career, and navigating a big company is as much of a skill set as any other, but this query also flagged up my experience in subscription or membership-focused organisations, and those with multiple brands and/or markets. They hadn’t occurred to me.

This can give you a starting point for company research, and filters for job alerts.

Then, once there’s a job you’re interested in, you grab the description, paste it into your LLM of choice and ask it a question to make sure it’s read the text – something like: What are the three key strengths someone would need to do well in this role?

Then add in your PII-free CV and ask it to connect the dots: What elements of this CV match well with the requirements in the job description, and where are the weaknesses or gaps?

Then you can reorganise your CV to highlight the most relevant skills and achievements, and fill the gaps (or use your cover letter to address them). For pity’s sake don’t use the LLM to produce your final CV or cover letter. Always refine with your human voice. Ask anyone who’s had to read through a mountain of job applications recently. They’re all AI generated, they’re generic as hell, and the rare human voice among them stands out in the best way.

Then, use it for interview prep: You are a <insert hiring manager’s role>, recruiting for the attached/following job <add job description>. What are 10 questions you are most likely to ask an applicant?

You can further refine this by once again adding the PII-free CV and asking what it would ask this specific candidate.

Job Applications – Recruiter Side

The most obvious use here, at least initially, is in the creation of the job description. There’s been a notable increase in the JD’s showing up on UK job boards with US spelling…

That aside though, they can be seriously good at it – provided you have a halfway decent idea of the type of employee you’re looking for, and being realistic about it (as in not trying to recruit an entire digital marketing team in one junior exec).

I’m hiring a part-time bookkeeper for a small e-commerce business with 3 employees. We’re friendly and informal, fully remote, and use Xero and Shopify. Write a job description including responsibilities, must-have skills, and a short ‘why work with us’ section. Tone: warm but professional.

The above obviously works better if you give it more context on who you are and why someone might want to work with you, but the actual responsibilities of a bookkeeper? You might want to see what the LLM gives you as a starting point.

Watch for biased language; sexist, racist, ageist, etc. It’s out there in current online job descriptions, which the LLMs will use for input so take the time to review your results for that, as well as an accurate description of what/who you’re after (as always).

Then screening, use the above JD + CV combination with a prompt to help you get some sort of rating on each application: Here is my job description: <paste/attach JD>. Here is a candidate’s CV: <paste/attach CV>. Rate the candidate’s fit out of 10 for each key requirement and give a one-paragraph summary of strengths and gaps. Do not make assumptions about anything not stated in the CV.

And since none of us ever get any training on how to interview, how about some suggested questions? Based on this job description <paste>, come up with 10 useful interview questions. Include a mix of competency-based, situational, and culture-fit questions. For each question, add a note on what a strong answer might include.

Yes, it does feel like we’ll be leaving the whole performance to the AI tools at some point, but in the meantime, a bit of preparation on both sides helps get better matches (hopefully).

Performance Reviews

Then, once they, or you, are in a job; you need to deal with performance reviews (insert pained grimace). As with the job application process, AI can support both sides of this, just make sure you’re using a secure instance of your LLM of choice to keep confidential data exactly that.

Remember too, the quality of your output is going to be very closely tied to the quality of your input. If you have no notes or record of performance to feed the tool, you’re not going to get much back, so think ahead.

I need to write a report for my mid-year performance review. Here are my KPIs, goals, and notes: <paste notes/weekly reports/previous review with this period’s goals>. Write a structured report with sections for achievements, areas for development, and suggested goals for the next six months. Tone should be constructive and specific, not vague or overly corporate.

Training

You can also use AI tools to help improve your knowledge and skills in a range of areas. Whether it’s by asking ChatGPT to create a 10-session training course on vibe coding for you, using the Quiz option in Notebook LM to make getting through key learning documents more interesting, or switching to the Learn option in Claude or Gemini rather than just grabbing the answers, you can use these tools to make you smarter, not the other way around.