Snowflake® wire-compatibility
Validated across data types, query syntax, aggregates, window functions, and semi-structured data — and broader compatibility beyond SQL dialect.
How we test →Why now
Human-centric B2B SaaS is at odds with agentic AI — on pricing, sandboxing, interfaces, and data access. Agents can’t pay the ~20x ops markup baked into SaaS compute. They can’t operate inside UI-first products. They can’t run safe dry-runs and evals at scale.
Fueled by the rise of agentic coding, SaaS repatriation is set to disrupt the $400B B2B SaaS market over the next five years. This wasn’t feasible 2 years ago — agentic coding crossed the threshold in late 2025.
What we’re building
Embucket is a managed SaaS repatriation platform — a secure home for bespoke rewrites of third-party SaaS, hosted on the customer’s enterprise lakehouse.
The initial product is a BYOC drop-in for Snowflake® workloads: wire-compatible, dry-runs supported, side-by-side operation on the same Iceberg tables in AWS S3 Table Buckets. Same data, your choice of engine. Deployable as a Snowflake® Native App.
18 of 22 queries pass
TPC-H SF100 — all 22 queries run in parallel on a 10 GB AWS Lambda. 191s wall clock, 600M+ lineitem rows, 865M+ total rows.
TPC-H details →Team
Founder
Engineering
Backed by
