Kleiner Perkins Grit Podcast

Frontier models still miss basic reasoning. Garrett Lord explains why expert feedback is becoming critical infrastructure for AI, and how Handshake’s elaborate network is helping labs build smarter systems. Tune into the full episode now.

I really enjoyed talking with Joubin Mirzadegan, of Kleiner Perkins, on his Kleiner Perkins Grit Podcast. Lots of fun Tripadvisor stories, the not-untypical entrepreneurial journey of near death before finding traction... and more than a few comments/lessons/stories of scaling a business. A nice "trip" down memory lane....

Before Plaid powered the fintech ecosystem, it started as a budgeting tool—with playbacks of your spending and push alerts about overpriced coffee. On Grit, Zachary Perret shares the missteps and half-built features that came first, and how a string of user drop-offs led them to uncover the real opportunity.

Before it became a platform serving over 460 million travelers a month, Tripadvisor began with one frustrating experience with travel agents. Steve Kaufer, co-founder of TripAdvisor, joins Joubin Mirzadegan to share how the company went from a struggling B2B startup to a consumer reviews platform that reached profitability in under a year. That journey led to a $200 million acquisition by IAC (later Expedia) in 2004 and a new era in online travel.

Handshake, best known as a student job platform, entered the AI training market in January. Just six months in, it expects to reach a $100M run rate from this new business line by the end of the year. The shift began last year when Handshake noticed that companies like Scale AI were actively using its platform to seek out PhDs in physics, law, medicine, and software engineering. But these hires weren’t for academia or traditional roles. They were being recruited to train LLMs. With a talent network that includes: → More than 500,000 PhDs → Over 3 million master’s students → 18 million verified members across 1,600 universities Handshake realized it had exactly what the AI labs were looking for: trusted access to highly specialized experts. This was not crowdsourced gig work, but real domain expertise. In January, the company quietly launched Handshake AI, a new division that works directly with foundation model companies. Experts are paid $30-160+ per hour to deliver high-quality human feedback in areas such as software engineering, finance, law, education, pharmaceutical research, music theory, and English literature. This is not a pivot. It is a second engine built on Handshake’s original mission. Garrett Lord, Handshake’s founder, discussed the expansion on Kleiner Perkins Grit Podcast on Monday. He noted that inbound demand from AI labs tripled overnight after Meta’s $14.3B investment in Scale AI last month. Google reportedly halted multiple projects with Scale. OpenAI and xAI paused some Scale projects, too, according to Scale contractors working on them. Handshake AI is now hiring quickly, adding around 12 engineers, operators, product managers, and researchers each week in SF and NY. While Scale remains a dominant player (it was reportedly on track to do $2B in revenue this year), the competitive landscape is shifting. Handshake is not the only workforce platform now expanding into AI data labeling. A growing number of workforce and gig-based platforms are now using their access to flexible, distributed talent as a competitive advantage, including: 🔹Mercor (an AI recruiting startup founded by two 21-year-olds, raised an $100M Series B at a $2B valuation this year) 🔹Uber (they're recruiting gig workers to complete projects for Uber’s internal business units but are also serving outside customers, including self-driving vehicle company Aurora Innovation and video game developer Niantic) Other AI data labeling players include: 🔹 Surge AI (reportedly booked over $1B in revenue last year as a bootstrapped business) 🔹 Snorkel AI (raised $100M in Series D funding at a $1.3B valuation in May) 🔹 Turing (raised $111M at a $2.2B valuation in March) As well as Invisible, Toloka, CloudFactory, Appen, Prolific, and Labelbox. What began as a student career platform is quickly becoming part of the infrastructure powering the next generation of AI. Sharing the full interview with Garrett here: https://xmrrwallet.com/cmx.plnkd.in/dxmNim8P

Garrett Lord, co-founder of Handshake, logged every referral and personally chased down the two names that topped the list. Those hires went on to become ‘iconic’, now core to Handshake’s growth.

Plaid started by helping users connect their bank accounts to apps. But founder Zachary Perret says that was just the beginning. In conversation with Joubin Mirzadegan on Grit, Zach shares how the company’s focus has shifted to solving systemic issues in financial services, using data and analytics in ways the industry hasn’t seen before.

When you build a platform that helps millions launch careers, the rise of AI isn’t a threat, but an opportunity. For Garrett Lord, co-founder of Handshake, it’s a chance to open even more doors. He joins Joubin Mirzadegan and Mamoon Hamid to explore how Handshake was built to give every student an equal shot, and how they’re now using AI to better match talent with opportunity at scale.

When Visa’s $5.3B acquisition of Plaid fell through, it could’ve been a setback. In this episode of Grit, Zachary Perret reflects on the unexpected chain of events that followed—and how staying independent turned out to be the best-case scenario.

To most founders, 5% monthly growth would be a win. To Daragh Murphy, it’s unacceptable. In this episode of Grit, he opens up about the emotional whiplash of scaling fast—why triple-digit growth still isn’t enough, and how he’s pushing for more.