T-5 Days to Prompt. Make sure to pack all your essentials 🧳

Building is the easy part in today's age but the real question is whether you want to own one long term. Ask yourself (and your team) these few questions first: 👉 Is the workflow truly unique to your business? 👉 Will you have engineering time to keep investing after the prototype? 👉 Are you comfortable with part of your data team becoming the team that maintains it? If most of those answers are no, the parts worth owning are your metric definitions, business logic, and the institutional knowledge that makes an agent useful. Not all the infrastructure around it. New on the blog from Nick Pasquinzo - link in the comments!

Today we announced that Greenflash is joining Hex. Excited to welcome the Greenflash team, and have them join us on building the best platform for data work in the AI era. Hear more from our CEO, @barrald → hex.tech/blog/hex-greenflash…

"Should I replace my BI tool with an agent?" Not exactly. Here are 5 🔑 takeaways from our new guide that explain why: 1. Analytics has split in two: AI tools nobody governs, and BI tools nobody uses. 2. A two-week wait for an answer is now a 30-second expectation. 3. The new BI test: can more people self-serve, can you trust the answer, and does it work where your team already works? 4. Good answers depend on context that stays current. 5. Bolting an agent onto your BI tool just means maintaining logic in two places. Grab the guide from the comments 👇

How to get on the Hex design team's good side? 🍦

ICYMI: @claudeai Dashboards connect directly to Hex. With this connection, you can ask your data a question in Claude, get an answer, and send it to Hex, while leveraging all of our rich context, collaboration, and governance capabilities. Read more: claude.com/resources/article…

Ten months ago, Account 360 felt like the future. Our customer team built it so anyone could check on any account, any time. Then the way the team worked with data changed, and the bar for "good" moved with it. Every new question became another tab, and the answers often showed up after the call they were needed for. Now the team asks Threads directly, and all the data behind Account 360 is what powers those answers. Nisha Dwivedi on what changed, and what she'd tell any customer team building something similar 👇

Watch it here: youtu.be/SuF2MHeipI0?si=qvS8…

Spark was everywhere. @databricks was making less than $1M a year. @pwendell, co-founder and VP of Engineering at Databricks, says the company had a few rough years where the open source project was taking off and the business wasn't. So they built something you could only get from Databricks. A managed service is, by definition, something you can't open source. And by the time it existed, Spark had already done the hard part of getting people to care. Open source wasn't the product. It was the top of the funnel. Full episode of Friends of Data on YouTube, link in the comments 👇