What’s one way you’ve used AI to simplify a task? Here's an example we've built
By George Barnett Reid 👀 · September 30, 2025 · Curated by George's Blog
What’s one way you’ve used AI to simplify a task? Here's an example we've built out for stage 1 of our amazon ppc candidate analysis.
We built out a custom form using Lovable, just to process the CV, then we use Cloudflare in the backend to process that CV and turn it to OpenAI in order to analyse the CV against our 'long' job description.
It then sends back a CV score out of 100, based on a series of 'must have' factors which helps qualify.
It would be impossible for us to process the number of resumes we get each week from Amazon ad managers. But we need a time effective solution to review these resumes and help identify where the seniority level lies.
Because ultimately we only want senior amazon ppc managers getting invited through to the next 2 video rounds. we don't want to occupy our time reviewing candidates that are never going to fit our model.
you might also notice we have a trick question in here. a tripwire as i like to call it.
people just dont read job descriptions. they're trying to smash out hundreds of applications and initially they dont really care about the job.
which is immensely annoying if you're reviewing them. just a time drain.
so our CV analysis removes most.
But then our tripwire also removes a chunk.
It's really easy to copy yourself. Just ask them to pick a number between 1 and 10. But in the job description, tell them the right answer is 23.
Its stupidly simple but it works. Most people arent paying attention. Do we reject all these people, no. Because sometimes we dont wanna judge too much. Ultimately in countries like India there is such vast competition they do approach it with a volume strategy. I don't want to remove high quality talent, at least not every time. Sometimes I'm more strict.
#itsalwaysdayone
If you want to remove 3 steps of your own hiring process, and like the idea of us delivering you a handful of candidates rather than reviewing 2000 (like we have to do), then check us out -> Pare.