What We Learned Between Design Preview and Public Beta

A few weeks ago, we announced Golden and opened a design preview. The response exceeded our expectations. Nearly a thousand companies requested access.

Rather than opening the doors to everyone immediately, we intentionally kept the preview small. We wanted to watch people use the product, understand where they got stuck, learn what mattered most, and continue investing in the infrastructure required to support customers at scale.

As we move into public beta, I wanted to share a few things we’ve learned along the way.

1. Enterprise demand arrived much earlier than expected

One of the biggest surprises was who showed up.

Among the customers we brought into the preview, roughly one in six came from Fortune 500 companies.

We expected interest from startups, data teams, and individual analysts. We did not expect such strong demand from large enterprises so early in our journey.

That changed our priorities.

Topics such as governance, security, compliance, auditing, and administrative controls moved much higher on the roadmap than we originally anticipated. Large organizations want the speed and accessibility of modern AI-powered analytics, but they also need the controls and reliability required to operate at scale.

The good news is that these investments benefit everyone.

2. Data is always messier than you think

Every analytics company eventually learns this lesson.

No matter how clean your demo data looks, real-world data is always more complicated.

We’ve seen spreadsheets that would make experienced analysts pause. We’ve seen databases with years of accumulated edge cases. We’ve seen inconsistent naming conventions, nested structures, and every imaginable variation of JSON stored inside database fields.

It’s humbling.

At the same time, it’s also where some of the biggest opportunities exist. Many of the improvements we’ve made during the preview focused on helping customers work with messy, imperfect, real-world data instead of expecting their data to conform to idealized assumptions.

Getting data right is not easy but its critical to make analysis actually work in the real world.

3. Simplicity always wins

One of the most consistent pieces of feedback surprised me. Nobody thought Golden lacked functionality. In fact, the opposite was true. People were often amazed by how much was already available. They could connect to data, analyze it, visualize it, and communicate insights without needing a collection of separate tools. Many commented that it didn’t feel like a typical V1 product.

But we learned that capability can become a liability if it’s presented poorly.

In our earliest onboarding sessions, people occasionally felt overwhelmed by the number of available options. The challenge wasn’t that the product couldn’t do enough - it was that it could do too much at once.

As a result, we dramatically simplified the experience. Features that previously required exploration are now often a click away. In many cases, we removed choices rather than adding them.

One lesson we’ve learned repeatedly is that customers rarely ask for simplicity directly. They ask questions, hesitate, or get lost. It’s our job to identify the underlying issue and design the right solution.

4. Chat became far more important than we expected

When we built chat into Golden, we thought of it primarily as a way to ask questions about data.

That’s certainly happening, but what surprised us was how much broader the usage became.

Customers use chat to understand unfamiliar datasets. They use it to learn how to perform tasks. They use it to format results. They use it to explore ideas. Sometimes they use it as an analytics interface. Other times they use it as product documentation, an onboarding guide, or a thought partner.

What we’ve learned is that users don’t think in terms of product boundaries. They don’t distinguish between “analysis,” “education,” and “workflow.” They simply want to accomplish their goal.

That insight has significantly influenced how we’re thinking about the future of the product.

5. Modern doesn’t replace the basics

One lesson that made me smile is how often customers asked for capabilities that have existed for decades.

People want to export to Excel.

People want to export to PowerPoint.

People want to inspect the underlying SQL.

People want to understand exactly how an answer was generated.

The newest technology doesn’t eliminate these needs. If anything, it makes them more important.

Trust matters. Interoperability matters. Existing workflows matter.

When we announced Golden, many of these capabilities didn’t exist yet. They do now, because customers made it clear that innovation isn’t about replacing everything that came before. It’s about making familiar workflows dramatically better.

6. There is so much more to building product than the features you see

The visible parts of a product get the attention.

The invisible parts determine whether it succeeds.

As we’ve prepared for public beta, we’ve spent enormous amounts of time on the things customers may never directly see: reliability, resiliency, backup and recovery procedures, security reviews, monitoring, logging, operational processes, and support systems.

These aren’t the features that show up in a demo.

They’re the foundation that allows us to move quickly while maintaining trust.

As more customers join the platform, that foundation becomes increasingly important.

The road to public beta

The most rewarding part of this journey has been seeing the progress.

In our earliest onboarding sessions, there were bugs in nearly every meeting. It was frustrating for our customers and frustrating for us.

Today, customers routinely connect data, analyze it, build insights, and communicate results with far fewer issues. The quality of the product has improved dramatically in a short period of time.

That progress wouldn’t have been possible without our early adopters. Their feedback shaped the product in countless ways. It also wouldn’t have been possible without a team that is deeply customer-focused and relentlessly committed to improving the experience every day.

As we enter public beta, we’re moving from a model built around one-on-one onboarding sessions to one designed to support thousands of customers.

But we’re still at the beginning.

The pace of innovation isn’t slowing down. If anything, it’s accelerating. We have a long list of capabilities we’re excited to build, many of which haven’t been publicly discussed yet.

The design preview taught us a tremendous amount. Public beta will teach us even more.

And we can’t wait to see what customers do next.