AI agents are all the rage right now, and early-stage startup Tezi is working on developing an AI agent to help HR teams find the best candidates for job openings. According to the startup, the bot will comb through resumes to find people who fit the hiring criteria, find interview times on recruiters' calendars, and email candidates.
Today, the company announced it has raised $9 million in seed funding to support its path to mass commercialization.
For now, the company is rolling out an alpha product to a small number of design clients this week, but that's the company's vision, said CEO and co-founder Raghavendra Prabhu, who acknowledged that HR departments have been using automated resume screening for some time, but Tej saw an opportunity to leverage a new generation of large language models (LLMs) to build more sophisticated recruiting tools for HR.
“We feel that the combination of reasoning and natural language gives us the option to build something completely different from what software has done in this space before,” he said.
Co-founder and COO Jason James believes existing tools aren't enough. “Let's say you get 1,000 applications for a job. In the past, AI, ML, algorithms have been good at determining that these resumes are really good,” he says. “But you still need a human to send emails, schedule interviews, etc. And now we're seeing end-to-end workflows, not just basic ranking.”
The founders acknowledge that for now, humans will need to remain involved in the process, and they expect it to become fully automated as the model improves. Additionally, the pool of candidates that emerges from job searches will depend on the quality of the prompts and job descriptions.
“They understand that automation can introduce bias, but they are working to mitigate it as much as possible. From their perspective, they take all the input from recruiters and objectively evaluate it against resumes. They say they can't control what the input is, but they try to minimize bias on their end.”
“Assuming bias is coming from employers, right now we're not doing enough to prevent that. What we do is make sure that by algorithmic means we're not letting in any kind of bias,” James said. They avoid looking at past hiring patterns. They want their models to match skills and other criteria set by hiring managers.
They trained their models on 250 million profiles they licensed from data providers and have worked with models from OpenAI and Anthropic to tune them to their hiring requirements.
The company is just getting started. It was founded earlier this year. They've started working with 15-20 design clients and hope to have all the kinks worked out and be able to distribute the beta more widely later this year.
The $9 million seed round was led by 8VC and Audacious Ventures, with participation from Liquid 2, Afore, Prime Set, South Park Commons and industry angel investors.