What can an AI interviewer do? And importantly, what can it not?

Posted by

Around this time last year, we made a decision that we had to build with Generative AI. Over the course of multiple iterations of user research, prototyping, and user tests, we identified specifically what we would build.

We will build AI agents that work with recruitment, tech teams, and candidates, helping them achieve great hiring outcomes. Our agents will be autonomous, operating at a reasonable level of human cognition, and will provide end-to-end outcomes.

We identified 3 agents to start with:

  1. Recruiter Agent – Scans resumes, matches them to JDs, and conducts screening calls with candidates.
  2. Tech Interview Agent – A fully customizable interviewer that can run first-round tech interviews in either a code-pairing mode or a theoretical mode.
  3. Candidate Agent – Helps candidates build adaptive resumes that showcase their match against the job and highlight their skills by taking self-evaluation tests.

You can see our AI agents here -> https://www.geektrust.com/assemble

Our team is small, so we had to prioritize our focus. Initially, we worked on all the above agents to gauge feasibility and direction. Once we grasped how to build with Gen AI and started talking to customers, we prioritized one of our agents to go to market with.

Which agent do we build first?

An insight most people miss is that tech teams’ time is crucial for fast hiring. You can have 100 recruiters presenting thousands of candidates, but if the tech folks don’t interview and say yes, you can’t make a single hire. The most valuable resource at any tech company is the tech team’s time. Considering the dependency on tech teams for hiring, shouldn’t we solve for this first? Also, which techie likes to do multiple interviews a day? So we decided to work on our AI tech interview agent.

What does our AI interview agent do?

I worked at Thoughtworks for 10 years. Our team knows how to build high-quality software products, and we’ve been helping companies hire developers, SDETs, DevOps, and data engineering folks for the last 10 years. We’ve placed thousands of techies over the past few years across companies ranging from early-stage startups to enterprises.

So we decided to focus on the space we know well—software developer hiring. The first version of our AI interview agent helps companies interview software developers. Our Java and Go interview agents are ready and going through pilots now. We’re building JS interview agents and will move on to cover C#, Ruby, and Python next. However, we have a customer who has asked for Frontend and SDET interviewers, so those will come into the pipeline quickly.

Who is our audience?

When we started out, I was clear that my target audience was services companies that hire throughout the year. But I have been surprised by the number of companies that have asked us to provide interviewers for them because they don’t want to spend their tech team’s time interviewing. Especially for the first round of interviews, where you tend to meet many candidates who are not a match. So our audience is any company where the tech team’s time is constrained and they need help running interviews.

I had a CEO of a $200 million IT services company ask me, “AI and all is fine, but what can you do for me?” I told him I could run 100 parallel tech interviews for him. Even with the size of his tech team, he couldn’t do that.

Recruitment teams are also very interested in our AI interview agent because it can provide additional value to their stakeholders. Imagine if you were a recruiter and had a techie capable of running tech interviews dedicated to your team. The quality of the profiles you give to your tech teams would be tremendous.

We believe we can reduce the time tech teams spend on interviews by 70%.

Once our other AI agents are in the picture, we believe we can improve time to offer by 4x.

What can AI not do? What are the drawbacks?

For most tech teams, culture is important. How the candidate responds to feedback is important, how well they listen is important, and, most critically, whether the team wants to work with this person is important. And AI cannot assess this. Sure, we will provide insight into how they respond to feedback, their articulation, etc., but whether you want to hire this person is a decision the tech team need to make.

We believe this can be assessed in the second round of tech interviews. And what our AI will do is make sure tech teams speak to matching candidates with the right amount of skill in the second round.

There’s more to share on how we built our AI agents, what’s the success we’re seeing, what issues are we facing etc. But that’s for another post :).

Currently, we’re running limited free pilots to customers. If you’d like a demo, hit us up. Write in at sales@geektrust.in

Leave a Reply

Your email address will not be published.