How are steps isolated?

Each workflow step runs in its own container on a machine slot. One machine runs one job at a time. Logs and artifacts stay after teardown.

Can I run on my own infrastructure?

Yes. Self-hosted fleets are an enterprise option. One install script enrolls a host; the host dials out, so nothing opens inbound. Get a demo about enterprise fleets.

What are distributed nodes?

A distributed node fans a FILE or FOLDER input into parallel jobs at run time. The parallelism budget caps how many machines run those jobs.

How do I watch a run?

Every step logs inputs and outputs. Artifacts and structured results stay queryable after the machines tear down.

Does scale mean rewriting the workflow?

No. The same graph runs on one machine or across a fleet. You change the parallelism budget, not the canvas.

Execution

The workhorse for the workflows

Set a parallelism budget. Managed pools provision machines for the run; self-hosted hosts dial in when online.

Get a demo

Run a workflow

Machines for the run. Then they tear down.

Pick a fleet in the run sidebar. Managed pools scale to the budget; enrolled hosts take the work.

Managed

autoscaled pool

Trickest provisions machines for the run, then tears them down.

Self-hosted

your hosts

Enroll hosts you own. One install script; the host dials out.

Both fleets

per-run choice

Keep managed and self-hosted. Pick one per run in the sidebar.

illustrative · pool count is a capability envelope

Same graph. More machines.

Parallelism caps concurrent machines for a run. Distribute shards a FILE or FOLDER into parallel jobs.

Parallelism budget

Run-time machine cap. Limits concurrent slots the run may hold.

Distribute

Graph config that shards a FILE or FOLDER into parallel jobs.

One job per machine

A machine runs one job at a time. Width comes from more machines.

Submit. Machines. Results.

A run locks a fleet and a parallelism budget. Machines take the work, then tear down.

Submit

The graph enters the queue with a fleet and a parallelism budget.

02live

Provision

Managed pools scale to the budget. Self-hosted hosts take jobs when online.

Execute

Each job fills one machine slot. Steps run in their own containers.

Tear down

Machines return to the pool. The run keeps its logs and files.

workflow.run · batch-scan

0done1running3queued

Schedule
queue
0.4s
runtime
running
Provision
fleet
1.2s
runtime
queued
Execute
nodes
live
runtime
queued
Persist
tables
0.8s
runtime
queued

replay · deterministicelapsed 0.0s

illustrative · timing labels are compositional

One container per step.

Each workflow step runs in its own container on a machine slot. Logs and artifacts persist after teardown.

Containerized steps

Every workflow step runs in its own container on a machine slot.

Logs per step

Stdout and stderr stay attached to each step after the container exits.

Artifacts that survive

Files and structured tables remain after machines tear down.

Stop and retry

Stop a run to free machines. Re-run a step without redrawing the graph.

questions

Frequently asked

Need capacity on your own hosts? Get a demo.

On managed machines Trickest operates, on hosts you enroll, or both. Pick a fleet in the run sidebar and set the parallelism budget.

Explore the platform

Workflows, schedules, the CLI, and fleets submit the same runs. Pick the next surface.

Get a personalized demo

See Trickest in Action

A 30-minute walkthrough. We map the platform to your stack and answer pricing and deployment questions for your environment.

Book a demo Explore the platform