AGENT WORKLOADS

Built for agents, from the engine up.

An engine that starts within a second. Sessions your agent connects to securely. Any agent, any model.

Try it out Talk to an engineer →

Sail CLI

Run Spark with one command.

sail spark run starts Sail on your machine, runs the script, and stops. Your agent gets a Spark engine whenever it needs one.

  • Runs on your laptop or in CI.
  • Runs your Spark SQL and DataFrame code.
  • Open source, with a sample skill for your agent.
Agent skill guide

Sessions

A live Spark engine your agent connects to.

A session is a Sail engine on your LakeSail cluster. Your agent connects securely, over an encrypted link that expires.

Sessions guide Read the Sessions post

  1. Create a session

    From the console or the API.

  2. Connect your agent

    A secure tunnel for this session only. The link expires.

  3. Query your tables

    The proxy runs in your cluster. LakeSail is never in your data’s path.

Headless

Build and manage pipelines without the UI.

Bring your own agent and model. With an API token, it uses the LakeSail API to create jobs, run them, and watch them.

API reference
Your agent
On any model you choose
LakeSail API
with your access
Create a job
POST/workloads/jobs
Start a run
POST/workloads/jobs/{job}/runs
Check its status
GET…/runs/{jobRun}
Stream its logs
POST…/runs/{jobRun}/logs/stream
Get notified
job_run.status.failed
job_run.status.succeeded

Getting started

Start on your laptop. Run on LakeSail.

  1. Try it on your laptop

    Install pysail and a Spark client, then add the sample skill.

    Installed

  2. Create a LakeSail cluster

    Sign up, connect your environment, and create a cluster.

    Ready

  3. Connect your agent

    Open a session for queries, and create an API token for jobs.

    Connected
Guides Agent skill guide Connect an external agent API reference

Give your agent a data engine.

Connect your own agent and model to LakeSail.

Try it out Talk to an engineer →

Privacy

Loading your saved choices…

Always on