Quick start
Create a Ploid credential and run your first agent research task.
1. Create an account
Create a Ploid account, then create an API key in Workspace settings → API keys. A secret is shown only once, so save it in your password manager or secret store.
export PLOID_API_KEY="ploid_live_..."2. Run a research task
curl https://api.ploid.com/v1/agent \
-H "Authorization: Bearer $PLOID_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"operation": "ask",
"prompt": "Research Manny Hernandez at Ploid. Confirm current identity and role, then return three non-creepy personalization angles with source URLs. Separate facts from inference.",
"max_acu": 8,
"response_format": "standard"
}'The API returns 202 with a durable run:
{
"data": {
"run_id": "run_123",
"status": "queued",
"poll_url": "/v1/agent/runs/run_123"
},
"meta": {
"acu_limit": 8,
"acu_used": 0
}
}Poll GET /v1/agent/runs/run_123 for the completed synthesis, typed artifacts,
and actual usage.
3. Decide how the result will be consumed
- Poll the durable JSON run for prose that another model or a person will read.
- Add an
output_schemawhen code must receive exact fields. - Use
Accept: text/event-streamfor long-running research and progress updates. - Use People Sets when the main deliverable is a durable list rather than a narrative answer.
4. Continue the task
Pass the same session_id when a follow-up should continue the same logical conversation.
{
"prompt": "Now rank those companies by stage fit and explain every exclusion.",
"session_id": "research_retool_01",
"max_acu": 12
}Long-term memory is off by default. Set memory to read or read_write only when the workflow explicitly needs it.