Platform
One platform, with the status of each part in plain sight.
We say what exists and what doesn't on the page, not in the small print.
● available · alpha · open source
Data: your lake in one container
- Catalog from the Delta log: schema, version, statistics and history.
- Read-only SQL over the tables you cite, with limits and Arrow streaming.
- Join without SQL: Periplo proposes the key, checks the relation in your data (1:N, N:1, N:M) before running and shows the SQL it runs.
- One container, one port, no database of its own.
docker run --rm -p 127.0.0.1:8080:8080 ghcr.io/massivedatascope/periplo:latest
● early access · Periplo Cloud
Teams and security
- Sign-in with your OIDC identity provider, members, invitations and service accounts.
- Grants with protection against privilege escalation, and an audit trail.
- Per-client isolation: separate bucket and key, temporary credentials, separate data planes.
- Row policies and redaction on a remote query engine.
○ coming next
Pipelines: follow and operate your ETL
See the runs that write your tables, launch them, and pause or resume their schedules, next to the data they produce. This part is in development and not in the current release. Your pipelines can already be built with loom today.
○ coming next · powered by loom
Build: from idea to a running service
Describe a service, get a loom project with senior practices already in place, deploy it to your AWS and observe it in Periplo. Until then, loom works on its own and our team can deliver your first loom project with you.
○ coming next
AI that follows your rules
An agent that builds with loom's patterns through typed changes you review. Today loom already runs agents defined in YAML, validated offline, with typed permissions and no secrets in the agent.
Polars by default, Spark when it pays
A loom pipeline is declared once and the engine is chosen in configuration. Python transformations are written for one engine; SQL steps are the most portable. Polars can read Unity Catalog.