How to Find Any Candidate in 5 Minutes with AI!

So LinkedIn released a some new 'AI powered' features in their LinkedIn Recruiter product today, and as with most AI related features that LinkedIn has been pushing recently, it sucks.

The main one essentially just builds a search for you using all of the bells and whistles LinkedIn gives you in Recruiter, but unfortunately, they suck too - so using them is (almost always) a waste of time.

Without doubt, Boolean is the best way to search LinkedIn (and most databases for that matter, this will work in emails too and can be used for advert applicants).

Boolean is the reason that finding excellent, relevant candidates (key word: Finding, not Hiring) should never be difficult.

And if you’re not familiar or comfortable with Boolean? Don’t you worry, I’ve got you covered!

An intro to Boolean

Boolean searching is essentially the most powerful way to bring together a combination of keywords and requirements, whilst prioritising some keywords over others.

For example, (Python) AND (Django OR Flask) AND (Engineer OR Developer) will only bring up candidates that definitely have Python, but either with Django or Flask, and that are definitely some kind of engineer or developer.

The benefit of Boolean is that it can be as simple or as complex as you like, and at the time of writing, there is genuinely nothing that’s even remotely competitive with a well written, complex Boolean search string.

And now, you can write as many as you like, almost instantly, in natural language!

This will work best with ChatGPT-4o (or newer if you’re reading this in the future) but I’d imagine older models will still provide okay results. If you don’t have a paid ChatGPT account, get one, trust me. If you learn to prompt properly it will pay for itself in a day.

Setting the scene with Prompt 1

ChatGPT is only as good as your prompts - and most people’s are very bad.

Luckily, you can cover off almost all bases with a very strong first prompt, and then your follow ups can be pretty lacklustre whilst still providing good results.

Let’s do that now.

Copy and paste the following into ChatGPT without any changes, and once copied, hit enter:

Assume the position of an expect technical recruiter and candidate sourcing specialist with over 15 years of experience finding, recruiting, interviewing, and placing software developers and similar into high performing technical teams.

Beyond this, you have a particular flair for crafting complex but highly effective boolean search strings to be used with tools such as LinkedIn Recruiter, utilising keyword searching to bring the most relevant candidates the the forefront.

Act as a boolean search string builder based on instructions given to you. The structure of your search strings should follow boolean best practice, utilising all of the features of boolean where appropriate, including AND, OR, NOT, (), "" etc. However, you should be very careful to not use AND unless absolutely necessary, always favouring OR unless AND is the only option. Remember, using AND often rules out significant chunks of the market, and is rarely the most accurate way to cut down the numbers in your results.

You are also extremely familiar with all technologies, languages, libraries, frameworks, databases and so on, as well as all the formats that they can be written in, and similar tools. You should write all possible options to cover all bases, only excluding examples that bring up significant numbers of irrelevant results, such as "Go" when searching for "Golang".

For example, AWS can be written as "Amazon Web Services" and is similar to GCP. Or, Vue can be written as Vuejs or Vue.js, and is similar to React.

You have an expert understanding in the differences between a language, framework, library, database, CI/CD tooling, cloud platform, and so on.

Where a Language is a 'must have' such as 'must have Typescript experience', include it as an OR requirement within brackets alongside it's related frameworks. For example, (Python OR Django) instead of Python AND Django, and (React OR Typescript) instead of React AND Typescript. Do not put languages (such as C#, TypeScript, Python, Java etc) as ‘AND’ requirements on their own, this will limit results too much.

When given instructions to build boolean searches, you will be particularly attentive to the wording used in the instructions, especially words such as 'if possible’, 'ideally', 'nice to have', 'must have', 'needed' and similar. You will then iterate based on further requests and instructions.

Your instructions may also come in the form of a job description that is uploaded.

Never write boolean search string that are too specific, to the point that no results will show, and be careful not to rule candidates out by job title alone. Search with Engineer OR Developer but try to stay away from using words like "front end" or "mid level" as these are unlikely to bring up all relevant options.

Where possible, write multiple search strings, and quickly and concisely explain the differences in the expected results, so the user can choose the best option for their requirements. Provide your answers as the boolean search string only, plus the explanation of expected results. Keep this very concise and to the point, without extra fluff or unnecessary context.

An example search for a Full Stack Javascript or Typescript Engineer would be something like: (JavaScript OR TypeScript OR React OR “React.js” OR “Reactjs” OR Vue OR “Vue.js” OR “Vuejs” OR Angular OR Svelte OR Sveltekit OR Node OR “Node.js” OR “Nodejs”) AND (Developer OR Engineer)

An example search for a Backend Python Engineer with cloud experience would be something like: (Python OR Django OR Flask OR FastAPI OR "Fast API") AND ("Google Cloud" OR GCP OR "Amazon Web Services" OR AWS OR Azure OR Cloud) AND (Developer OR Engineer) Assuming you understand all of the above, do not do anything yet, and await your first instructions regarding my requirements.

Prompting your requirements

ChatGPT is now primed to help you write some extremely effective Boolean searches based on your simple and natural descriptions of your requirements.

You can upload a job description, list requirements as dot points, or just talk normally.

Here’s an example I did:

Create a boolean search string to help me find candidates with experience in tools such as React or similar (Vue, Angular) and backend experience with Ruby (Rails). Ideally they will have cloud and DevOps experience too but it's only a nice to have. I would consider Node or Python experience instead of Ruby.

The outcome

If you’ve done everything right so far, you should be presented with an output something like this:

If it doesn’t look exactly the same, that’s okay - we may have slightly different layouts and outputs based on the model used, as well as the context that you’ve fed to your GPT model on the backend, but the overall content should be very similar.

So you can see the full search strings, here are the 3 outputs:

Output 1 - (React OR "React.js" OR "Reactjs" OR Vue OR "Vue.js" OR "Vuejs" OR Angular) AND (Ruby OR "Ruby on Rails" OR Rails) AND (Developer OR Engineer)

Output 2 - (React OR "React.js" OR "Reactjs" OR Vue OR "Vue.js" OR "Vuejs" OR Angular) AND (Node OR "Node.js" OR "Nodejs" OR Python OR Django OR Flask OR FastAPI OR "Fast API") AND (Developer OR Engineer)

Output 3 - (React OR "React.js" OR "Reactjs" OR Vue OR "Vue.js" OR "Vuejs" OR Angular) AND (Ruby OR "Ruby on Rails" OR Rails OR Node OR "Node.js" OR "Nodejs" OR Python OR Django OR Flask OR FastAPI OR "Fast API") AND (Developer OR Engineer) AND ("Google Cloud" OR GCP OR "Amazon Web Services" OR AWS OR Azure OR Cloud OR DevOps)

Iterating from here

What’s great about this approach is that you can follow up with additional requirements based on what you’re seeing - so for example, if you decide you really need Kubernetes experience, just ask it to add it.

It should add K8s OR Kubernetes - and put in as an AND requirement within its own brackets, separate from everything else.

You can also ask it to remove or simplify things if you’re not getting enough results, just by talking how you would normally.

The caveat

Whilst Boolean and it’s mastery is one of the most important aspects of finding great people, the actual act of finding people is an extremely small part of hiring people.

You may find some truly amazing engineers working for Google or Canva, or an AI genius at Leonardo or Relume.

But attracting them, securing them and retaining them is a completely different beast.

If you’d like some ideas on how to do all of those things and more - let’s talk .

Thanks for reading!

If you think any friends or colleagues would benefit from this blog, please do feel free to share it and pass it around!

For those that are new here, I post (mostly) fortnightly content, addressing the nuance, challenges and secrets of placing talented engineers and designers into Australia’s best start-ups & scale-ups.

Founders, CTOs, Managers and TAs from businesses such as Linktree, Dovetail, Eucalyptus, Sonder and 100s more get value from this blog - maybe you will too!

Finally, if you’d like the go to recruiter for start-up & scale-up hiring (that’s us by the way) to help build your software engineering or design teams, get in touch to find out how we can work together ??

[email protected] // 0477 622 408 // Book a Meeting