ptn_demo_7f3a
Partition · N. Virginia · created just now
Services in this partition
Workflows
0 calls · 0.00 ACU
Spend by service
Ledger
- No usage yet.

Reliable multi-step work.
Describe a sequence of steps once in plain JSON or YAML and let the platform run it: branching, parallel steps, retries, timeouts, and pauses for human approval.
Declarative definitions
Write the steps, not the plumbing.
Retries and timeouts
Hard limits stop runaway loops and runaway spend.
Human in the loop
Pause a run until a person or another agent signs off.
Full history
Every run records each step, its input and output, so it can be audited and replayed.
{ "tool": "workflow.run", "name": "weekly-report", "input": { "week": 41 } }Interactive simulation
Press Run to send the call. The dashboard shows what it would use in a partition. Simulated: nothing is actually executed.
# Start a multi-step run.
→ POST /mcp
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "workflow.run",
"arguments": {
"name": "weekly-report",
"input": {
"week": 41
}
}
}
}# See every step recorded.
→ POST /mcp
{
"jsonrpc": "2.0",
"id": 2,
"method": "tools/call",
"params": {
"name": "workflow.history",
"arguments": {
"run": "wr_aa17"
}
}
}Ready. Press Run.
Partition · N. Virginia · created just now
Services in this partition
Workflows
0 calls · 0.00 ACU
Spend by service
Ledger
Why it matters
A multi-step agent job is easy to write once and hard to make reliable: retries, timeouts, approvals and restarts all need code.
Reliability
Retries with backoff and hard time limits on every step.
Control
Pause for approval from a person or another agent.
Audit
Full step history with inputs, outputs and cost.
Spend
Per-run credit limits and automatic trigger throttling.
Workflows run inside one partition. Very large fan-outs and visual editing are not included.
In depth
Chain Memory, Store, AI, QC, VM or any other tool into a workflow with a short JSON or YAML document.
Steps with no dependency on each other run side by side. You only declare what must wait.
Skip steps, choose between paths, or jump to a handler when something fails.
Run the same steps over each item in a list, with a concurrency limit you control.
Pause for a timer, until a time, or until a person or another agent approves.
Call one workflow from another to build bigger processes from reusable parts.
Retry with backoff on the errors you choose. Every step and every workflow has a hard time limit.
Short workflows return their result in the same call. Long ones return a run ID and can run for up to 7 days.
Start a workflow from an event on a stream, or on a schedule.
Every step’s input, output, cost and timing is recorded. Replay a run, or restart from any step using the recorded earlier results.
13 tools your agent can call. Each one is a standard MCP tool with a typed schema and a price shown in tools/list.
Safeguards against runaway spend: every run has a credit envelope, retries are capped, timeouts are mandatory, and a trigger that fires too fast is switched off automatically. If a run runs out of credits, it stops cleanly with a payment request, and you can top up and resume from that step.
Market research swarm
Many agents research competitors in parallel and merge their findings.
Due diligence assistant
Work through a data room and flag what needs a human.
News and signal monitoring
Watch sources continuously and alert on what matters.
Competitive pricing tracker
Track prices over time and surface meaningful changes.
Autonomous contract and document auditing
Audit hundreds of contracts, filings or codebases in parallel.
Automated code review
Review every change, run the tests, and report.
and 18 more