Interview

Julie Zhuo on building Sundial — the 'Cursor for data analysis' — and lessons from Meta's data-driven culture

Aug 21, 2025 with Julie Zhuo

Key Points

  • Sundial, co-founded by Julie Zhuo, automates data analysis work for companies scaling past product-market fit, with customers including OpenAI and Character.ai.
  • Zhuo applies Meta's data-driven playbook: behavioral data trumps stated preferences, as shown by the 2006 Newsfeed backlash that proved users valued the product despite vocal opposition.
  • The right time to adopt data infrastructure is during scaling, not early product development, when companies have proven demand and need to optimize what's already working.
Julie Zhuo on building Sundial — the 'Cursor for data analysis' — and lessons from Meta's data-driven culture

Summary

Julie Zhuo is a co-founder of Sundial, which she describes as the Cursor for data analysis. The product automates work data scientists do so companies can make better decisions faster. OpenAI and Character.ai are current customers. Before founding Sundial, Zhuo led design and research for the Facebook product at Meta.

Zhuo's pitch draws from what she watched Meta build. Meta's operational rigor—its ability to instrument, model, and optimize the business—was entirely homegrown. Zhuo credits that infrastructure with enabling the relentless experimentation cycle that drove its growth. The growth team concept itself, associated with early figures like Chamath Palihapitiya and Alex Schultz, was pioneered at Facebook when growth stalled and the company had to get systematic about understanding what was working.

Preference falsification and the Newsfeed lesson

The 2006 Newsfeed launch offers the clearest illustration. Within a week, a million users joined a Facebook group called "I F***ing Hate Facebook Newsfeed," convinced it was surveillance creep. Mark Zuckerberg and the team kept it anyway. The behavioral data told a different story. People were spending significant time on Newsfeed, and the very reason they found the protest group was through Newsfeed itself. The data overrode the stated preference.

Zhuo's framework from that experience treats qualitative feedback, customer surveys, and behavioral data as all equally data. Behavior—where people actually spend time and click—is the ground truth. The goal is to hold all of it in one place to build a realistic, influenceable model of how the business works.

When Sundial makes sense

Zhuo cautions that obsessing over data too early can be destructive. Before product-market fit, companies need to take bigger swings, not optimize. The right moment to engage Sundial is when a company already has product-market fit and is scaling, when it is starting to hire growth or data roles and needs to understand what is working well enough to put more fuel on the fire.

Laura Dada

Laura Dada, who appeared in the Wall Street Journal that week, joined Meta as a design lead. She took over Facebook Messenger when the team was around eight designers and scaled it to roughly 200 before becoming the product group lead and effectively the GM of Messenger. Zhuo notes that designers becoming GMs is rare at Meta. Adam Mosseri at Instagram is the only other example she can cite. Dada most recently moved to AI work at Meta and had just been announced as Figma's new Chief Design Officer.