Saving Babylist 25%: "We used to allocate 10 to 20% of engineering capacity to managing tech debt... With Espresso, almost all of that is gone"

Babylist and Snowflake

Andy Beyerly is a Senior Analytics Engineer on Babylist's data team, which runs on Snowflake.

"On the data side, my job is to make sure every team building those products and experiences has the trusted data they need to make great decisions," he says.

Every team at Babylist queries Snowflake. The cost and the runtime of all of it lands on Andy's team.

"Snowflake and Espresso act as a stable, reliable backbone for our entire data platform. They let us maximize the time we spend doing what we're actually good at, building great data products for our users, instead of maintaining infrastructure."

Espresso steps in

Babylist came to Espresso with organic growth and a team tired of hand-tuning warehouses to keep up with it.

Espresso replayed three months of Babylist's workload metadata through its simulation environment and estimated 21%. Setup was a five-minute script: they created a service user and pointed one connection URL at Espresso's proxy. They changed no workflows and rewrote no SQL.

Babylist had been paying for a separate Snowflake Enterprise account to get the horizontal scaling their BI tool needed. Espresso's multicluster at home replicates that on a Standard account, without the upgrade.

Savings are tracking to 25%, and Babylist's own independent analysis landed on the same number.

The results

"We're spending 25% less on Snowflake credits since implementing Espresso. That's bandwidth we've redirected toward expanding our product offering and driving the business forward."

"Realizing we could consolidate our entire Snowflake footprint into a single account using Espresso's warehouse clustering. Our jobs run faster, our costs are lower, but the bigger shift was cultural. The team stopped thinking about infrastructure as a constraint. We stopped having conversations about 'can we afford to run this' and started having conversations about 'what should we build next.'"

Scaling AI without scaling the bill

Babylist's flagship build on the reclaimed time is Data Doula, an internal AI data analyst running on Snowflake Cortex Agents over governed semantic views, so any business user can ask a natural-language question and get a metric-consistent answer without writing SQL. The semantic layer underneath it is where dashboards, embedded apps, and agentic experiences all read from, which means query volume climbs every time another team adopts it.

"Normally when you start scaling AI-driven data access, the compute bill becomes the conversation. With Espresso, that's not even a talking point for us, which means we can just focus on building."

"Espresso makes sure that as AI-driven query volume grows, our compute costs don't grow with it, which means budget and engineering capacity stay pointed at families, not infrastructure."

Want to know what Espresso would save you? The savings estimate is a 5-minute script, results come back in a few days, and there's no commitment. Email ben@espresso.ai.

‍