(Bloomberg) -- Rippling, a business software startup, is rolling out a bot-forward way to size up a new employee. The product, Talent Signal, will tap corporate willingness to experiment with different ways to use artificial intelligence.
Talent Signal aims to take bias out of evaluations by using task-specific metrics to evaluate how well an employee is performing — and then provide feedback, such as analyzing call transcripts and message exchanges to encourage a support agent to make sure they acknowledge customer emotions.
Rippling said the tool can identify patterns a manager might miss, helping employers better deploy resources. The goal is to identify when a worker might need fast-tracking or more guidance.
“In the ideal case, the manager would pull up a chair next to the employee, and spend 90 days with them,” coaching them, said Matt MacInnis, Rippling’s chief operating officer. Few managers have time for that, but AI does.
The San Francisco-based startup said customers who have tested the product say it provides a way to check their preconceived notions and look objectively as they evaluate an employee’s performance.
MacInnis acknowledged some people might find the tool creepy, but it’s no creepier than a manager keeping tabs on a new employee and much less biased.
The routine 90-day evaluation “can’t be about the vibe,” he said. “It’s got to be about the facts.”
The tool works by analyzing a set of projects, such as customer inquiries for a support agent or writing code for an engineer. It takes into account how complicated or simple the projects are, and assesses how the employee handled them.
At 90 days, Talent Signal assigns the employee one of three rankings: high potential, typical or pay attention. It provides specific examples, and allows a manager to agree or disagree with its findings by clicking a thumbs up/thumbs down icon and providing their reasons.
The software was built by Rippling engineers, who based it on AI models from OpenAI and Anthropic. For now it’s free to customers, with Rippling still determining if and how it might charge for Talent Signal in the future.
Rippling trained Talent Signal on data from its own workforce, plus some customers who opted in to early testing and development. It could eventually roll out to analyze performance beyond the initial 90-day period.
©2024 Bloomberg L.P.
