Why we’re building AI agents.
We stopped all active product development when chatGPT launched. For the first few days our (combined) jaws dropped at what it was capable of, and we pinched ourselves to check if we were dreaming.
Very quickly it dawned on us that GPT would kill our business.
At Geektrust we enable companies to hire technologists purely based on their skill (and not their “pedigree” or what their resume says). We built an AI product that could evaluate code and recommend candidates for interviewing based on the code they wrote.
But two things happened when chatGPT launched.
- Our ML code assessment engine that we painstakingly built over 4 years became commoditised. GPT could assess 10x better than us, could check for any tech stack, and was now available for all companies to use. We didn’t have an edge anymore.
- Candidates could now ask GPT to solve our coding challenges, and get a well crafted solution in seconds. Plagiarism would be rampant.
We whined about it for a few days but we realised that GenAI could be this magic wand that helps us truly bring out a candidate’s skill.
We started experimenting with GPT to see what we could do with it. If we had no constraints, how would we spot a candidate’s skill? The answer was fairly obvious — spend time with the candidate, solving a problem together, and discussing/debating solutions.
Could we actually build something like this out? Could we enable candidates to showcase skill in a safe, engaging environment and in a manner that companies would value? Could we replace 1st round technical interviews at companies?
That’s what we’ve been trying to find out over the last 6 months.
And now, finally, we have the 1st version of our AI code pairing interview agent ready for trial.
You can access the beta version 1.0 here -> https://www.geektrust.com/share/interview-agent
AI will help candidates and companies have better hiring outcomes.
There is no question about it. All we need to figure out are the exact mechanics of it. We’ve been trying to make hiring easier for 8 years. GenAI now gives us the super power to make it happen.
In addition to our interview agent, we’re building a recruiter agent and a candidate agent. These agents will embed with tech and recruitment teams to do the heavy lifting for them.
In our user testing, we got feedback from a developer that this was the least intimidating interview she’s ever done. No one was judging her all the time and she felt she could showcase her skills at her pace. This was the best feedback we could have hoped for. Maybe we can build a truly humane AI interviewer.
Write in to us here if want to connect or if you have feedback. Cheers!
