Warehouse Management

The control plane for the Snowflake credits you do still spend. Melt right-sizes warehouses, schedules suspends and resumes around your team’s actual hours, and enforces per-team budgets at admit-time — from the same proxy that already sees every statement.

What it’ll do

The capabilities, at a glance.

Continuous right-sizing

Melt watches a rolling window of statements through the proxy — bytes scanned, runtime, spill, queue depth — and sets each warehouse’s size for the next hour against the workload mix it actually saw.

Schedule-aware suspend and resume

Auto-suspend stays on as a safety net, but Melt layers a calendar over it: pre-warm the BI warehouse at 7:55am Mon–Fri so the first dashboard isn’t cold; suspend the dbt warehouse at 6pm and leave it down through the weekend.

Per-team and per-workload budgets, enforced at admit-time

Melt classifies every statement and tags it by team / role / query-tag / warehouse.

Built for

Where this fits.

The 4–15-person data platform team

Owns Snowflake spend at $200k–$2M/yr scale.

The multi-tenant SaaS data team

Bills customers based on Snowflake usage.

The platform engineer adding FinOps to an agent rollout

Just shipped agentic SQL into production, watched QUERY_HISTORY 5×, and got a Slack message from finance.

Want warehouse management in your stack early?

We’re shipping this with a small group of design partners. Tell us about your workload and we’ll set you up.