Built in Rows: The Sigma Newsletter with Katrina Menne

You’re probably overthinking visualizations.

We spent years living in tables that held all the answers but made patterns hard to spot, then swung to charts that looked great but left out the underlying details people need to act, and somewhere along the way we forgot that the real job of a visual in a business context is to point you toward what matters while keeping the details close enough to answer the next question without leaving the page.

That gap between signal and action is where most visuals break down, because they either impress without direction or overwhelm without clarity, and the result is the same follow-up email asking for the why behind the number. In my latest blog, I walk through how to design for the moment when someone can say, "I see what is happening, and I know what to do next, and how to choose the right level of visual for the audience in front of you without slowing the work down." If that sounds like the reset your team needs, read the full post.

— Katrina Menne

Khush Gill is turning AI into real workflow wins, with real dollar impact. In his new blog, he explains how he built “SE Buddy” inside Sigma to coach demos, prep next calls, and even draft POC plans. And the thread through it all?  Keep sensitive data in the warehouse, move fast, and let domain experts lead.

Katrina: Aside from the obvious reason of you working at Sigma, why did you choose to build this SE buddy in Sigma instead of somewhere else?

Khush: Two reasons. First, governance. We’re dealing with customer calls and internal enablement material, so we can’t send that to a public LLM. Sigma keeps all our data and queries in Snowflake, so it’s really the only safe place to do this. Second, context. A public model doesn’t know what a good SE at Sigma looks like, so I needed an easy way to add our domain knowledge, and Sigma makes it easy to build. 

Katrina: Speaking of easy, AI can feel intimidating to build with. How much technical depth do you think folks need to start?

Khush: Right, so it is intimidating, but with Sigma, it’s really not because we take care of the infrastructure. The people who have all the context in their head, like in this example of what a good SE or sales rep is, they’re not PhDs in LLMs; they’re PhDs in that specific business name. So then the obstacle becomes your creativity because in Sigma, the technical part is simple. All you need to do is connect a prompt to a text input box, select the transcript column, and then you see the output. 

Katrina: In your blog, you encourage folks to look for repeatable processes to build with AI. Do you have any advice for unleashing creativity when developing solutions?

Khush: Just start. Start with an empty workbook, spend 30 minutes building out a simple solution on some data, and start asking for feedback. It’s better to cultivate a culture around replacing repeatable workflows that get people really excited and solve 70-80% of the solution instead of waiting for 100%. Focus on rapid prototyping so the conversation can be about producing quick wins and continuously improving.   

Curious how it came together? Read his blog, The Day I Weaponized Our Data Warehouse, for the details. Then catch the walkthrough in From Prototype to Production: Building and Scaling AI Systems in Sigma

I'm not sure if I can express how excited I am for bulk row updates. By updating multiple rows at once, Sigma data apps are easier to build (especially when paired with if/else actions) and more powerful.

This video shows several scenarios and includes two short walkthroughs on how to build them with bulk row updates. I'd love to hear what solutions you come up with, too!

One of my favorite parts of the data community is how generous people are with real solutions and honest lessons learned. That openness dramatically helped shape my career. I learned faster because someone shared their workbook, their pitfalls, and the thinking behind their choices. I've seen previews of these sessions, and they keep that spirit alive! I'm super excited to learn from them and bring a few new patterns into my next build.

• Real-Time Headcount Planning for Finance Leaders — A practical dive into evaluating hiring scenarios and their financial impact in one place, so finance and ops can align faster.
• Leading with Data: The Data Analyst’s Path to Leadership — Data Leaders Luke Stanke and Joe Thomas share frameworks you can use right now to level up influence, not just outputs, and translate analysis into decisions.
• Building and Scaling AI Systems in Sigma — A practical session on LLMOps in practice, including a realistic 90-day roadmap to move from prototype to a trusted system that your team can operate, iterate on, and get meaningful value from.

So much good stuff in motion right now, with new features, sharper solutions, and cool learning opportunities, and I’m excited you’re along for the ride!

Thanks for tuning in, and I’ll see you in the next one 👋

-Katrina