No PriorsNo Priors Ep. 83 | With Rippling COO Matt MacInnis
Sarah Guo and Matt MacInnis on rippling’s Talent Signal Uses AI To Grade Employee Work, Not Vibes.
In this episode of No Priors, featuring Sarah Guo and Matt MacInnis, No Priors Ep. 83 | With Rippling COO Matt MacInnis explores rippling’s Talent Signal Uses AI To Grade Employee Work, Not Vibes Rippling COO Matt MacInnis discusses the company’s compound-startup strategy and unveils Talent Signal, an AI product that evaluates employees based on their actual work output rather than managerial impressions. The system ingests work product data from tools like GitHub and Salesforce, combines it with HRIS data such as role and level, and produces a calibrated signal on new hires after 90 days. MacInnis argues this can surface overlooked high performers, flag struggling employees earlier, and reduce biased, vibe-based performance reviews. He also emphasizes cautious rollout, strict internal policies against AI-only employment decisions, and openness to scrutiny around bias and ethical use.
Rippling’s Talent Signal Uses AI To Grade Employee Work, Not Vibes
Rippling COO Matt MacInnis discusses the company’s compound-startup strategy and unveils Talent Signal, an AI product that evaluates employees based on their actual work output rather than managerial impressions. The system ingests work product data from tools like GitHub and Salesforce, combines it with HRIS data such as role and level, and produces a calibrated signal on new hires after 90 days. MacInnis argues this can surface overlooked high performers, flag struggling employees earlier, and reduce biased, vibe-based performance reviews. He also emphasizes cautious rollout, strict internal policies against AI-only employment decisions, and openness to scrutiny around bias and ethical use.
Key Takeaways
Evaluate performance from work product, not manager ‘vibes’.
Talent Signal focuses solely on concrete outputs—code, sales interactions, support tickets—rather than demographics or subjective impressions, aiming to reduce bias and make reviews more fact-based.
Use AI as a signal, never as the sole decision-maker.
Rippling’s internal policies forbid relying exclusively on Talent Signal for promotions, terminations, or other employment decisions; managers must still perform holistic, human judgment and review underlying examples.
Start with constrained, low-scope deployments to build trust.
The system currently generates a single signal at the 90-day mark for new hires, allowing companies to back-test, assess accuracy, and expand usage gradually without overstepping the organizational ‘Overton window’.
Leverage unique platform data for defensible AI products.
Rippling’s advantage comes from unifying HRIS data with work systems (e. ...
Target roles and cultures already oriented around coaching and metrics.
Early versions focus on IC engineers, salespeople, and support agents—functions where output is traceable, coaching-heavy cultures exist, and leaders are hungry for competitive performance insights.
Use AI to surface hidden talent and support at-risk employees.
Examples from dogfooding show the model elevating under-the-radar high performers and flagging team members who need help ramping, enabling better coaching, fairer calibration, and stronger team performance.
Don’t chase generic AI features; build revenue-driving, unique products.
MacInnis criticizes me-too chatbots and copilots, arguing AI investments should go to capabilities that exploit a company’s ‘vibranium’—its distinctive data and platform strengths—to create new, monetizable SKUs.
Notable Quotes
“Where the magic really comes is where there's something common underneath all of these different applications that provides you with what I like to call your vibranium advantage.”
— Matt MacInnis
“If you wanna know if someone's a good engineer, look at their contributions. Like, look at their source code.”
— Matt MacInnis
“The motivating factor here, honestly, it's the bad manager… Talent Signal walks into that environment and slams your work product down on the table and says, 'What about this?'”
— Matt MacInnis
“Talent Signal is not making employment decisions. It's just giving this independent signal to the manager about how the employee is doing.”
— Matt MacInnis
“I'm thankful for the pitchforkers… when someone comes at us and asks hard questions about bias or unintended consequences, we're just gonna listen and we're gonna learn.”
— Matt MacInnis
Questions Answered in This Episode
How can companies validate that Talent Signal’s outputs are free of hidden bias, given that it’s trained on historical performance data that may itself be biased?
Rippling COO Matt MacInnis discusses the company’s compound-startup strategy and unveils Talent Signal, an AI product that evaluates employees based on their actual work output rather than managerial impressions. ...
What specific safeguards or audit mechanisms should be in place before expanding beyond a single 90-day signal to continuous performance monitoring?
How might employees’ behavior change once they know their work product is being systematically analyzed by AI, and could that create new kinds of gaming or pressure?
In what ways could Talent Signal misattribute performance in highly collaborative environments, and how should managers adjust their interpretation to account for that?
How should organizations communicate the use of AI in performance assessment to employees to maintain trust, transparency, and a sense of fairness?
EVERY SPOKEN WORD
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome