Hex

A data analyst at a mid-size company runs an analysis on why Q3 revenue dipped. It takes two days. The notebook disappears into a shared drive. Two months later, a different analyst runs the same analysis. The question was already answered - it just couldn't be found, shared, or trusted. This is not a rare story. It is the default state of data work at most companies.

Hex is the company that decided to fix this. Not with another dashboard tool. Not with a smarter chatbot. With a single, connected workspace where the analysis you run today becomes the foundation someone else builds on tomorrow.

Three People, One Frustration

Barry McCardel, Caitlin Colgrove, and Glen Takahashi all worked at Palantir. They were good at building data tools. They were also deeply familiar with what data tools consistently failed to do: connect the person doing the analysis to the person who needed the answer. The gap between "notebook full of insight" and "decision made by a VP" was routinely filled by screenshots, PowerPoint slides, and urgent Slack messages at 9pm.

In 2020, they left to build the tool they always wished existed. They incorporated Hex Technologies and started with a deceptively simple idea: what if a notebook wasn't just where you did the work, but also where you shared it, governed it, and let others explore it? What if writing SQL and building a dashboard were part of the same document?

The beauty of code, McCardel argued, is its infinite flexibility. The tools of the future should unleash that flexibility, not constrain it.

The Product Problem Nobody Admits

Most analytics tools are point solutions. Jupyter for notebooks. Looker for dashboards. Mode for SQL. Tableau for charts. Every tool does one thing well and creates a handoff problem when the work needs to move to the next stage. Data analysts have spent years building elaborate rituals around these handoffs: exporting CSVs, embedding screenshots in Notion docs, fielding Slack messages from colleagues who don't trust the numbers in the slide.

Hex's argument is that the problem is structural, not cosmetic. Switching tools mid-analysis doesn't just slow you down - it breaks the chain of provenance. You lose the connection between the raw data, the transformation logic, and the final chart. Someone somewhere has to trust a number that they can't trace back to its source.

In a world where AI is writing more of the code, that traceability matters even more. If an AI generates a SQL query and nobody can audit it, the resulting insight isn't trusted data - it's a confident guess. Hex's Context Studio and semantic layer exist precisely to solve this: data teams govern what the AI can access, define the metrics it works with, and ensure the answers that reach a VP came from sources the team actually endorses.

The Proof

The Companies That Bet On It

Hex's customer list reads like a Silicon Valley attendance sheet: Reddit, Figma, Anthropic, Brex, Ramp, Rivian, StubHub, HubSpot, Cisco, the NBA, Vercel, Cursor. Over 1,500 organizations in total. The name that stands out is Anthropic - the company that makes the AI models Hex uses to power its own agents is itself a Hex customer for data analysis. That is either a very good endorsement or a very good coincidence. Probably both.

StubHub used Hex to replace legacy tooling that was slowing down its data team. Mercor credited the platform with helping unlock $100M in revenue. A Figma analyst simply said: "I never have to build another chart again." The Brex data team reported saving 20 minutes per metric investigation, per day.

These are not edge cases. Hex's annual State of Data Teams report, which surveyed over 2,000 data professionals in 2025, found that 77% of data leaders are excited about AI possibilities - but only 3% said it was their main current focus. The gap between excitement and execution is exactly the market Hex is building into.

Why It Matters

Back to That Analyst's Notebook

Remember the analyst who ran a two-day analysis that nobody could find? In a company running Hex, that analysis lives in a shared workspace. It is versioned. The metrics are defined in a semantic model that the data team governs. A colleague can open a Thread, type "why did Q3 revenue dip?" in plain English, and get an answer that traces directly back to the original work - without running it again, without finding the notebook, without a 9pm Slack message.

The problem Hex is solving is not just analytical. It is organizational. Data insight is worth nothing if it stays in a notebook. The insight has to travel - from the analyst who found it, through the layer of governance that makes it trustworthy, to the decision-maker who needs it. Hex is building the infrastructure for that journey.

With $177M raised, 1,500 teams paying for the platform, and AI agents now capable of fielding natural language questions against a governed data layer, Hex has made a credible case that the journey is possible. The remaining question is whether they can own it at scale before Databricks, Google, and Snowflake decide to build the same thing directly into their own stacks.

McCardel's answer, essentially, is that integration is the moat. Anyone can build a chatbot that writes SQL. Building a workspace where that SQL lives next to the Python analysis, the semantic model, the published data app, and the governance layer - all in one document, all in one company - is the harder problem. Hex has a five-year head start on solving it.

That analyst's notebook is no longer missing. It is the foundation for the next ten analyses. That's the whole bet.