Orchestrate intelligence
at enterprise scale.

$ stelux swarm deploy research-desk \
    --agents planner,analyst:4,critic \
    --data lakebase://finance/filings \
    --compute local:h100x8 --burst cloud:a100 \
    --guardrails pii,budget:1200/day

→ validating manifest        research-desk.yaml
→ provisioning agents        3 roles · 6 replicas
→ attaching lakebase         finance/filings
→ warming gpu pool           local:h100x8
✓ swarm live  api.stelux.ai/v1/swarms/rd-8f2

from stelux import Swarm, Lakebase, Compute

swarm = Swarm("research-desk")
swarm.add("planner")
swarm.add("analyst", replicas=4)
swarm.add("critic", verifies="analyst")

swarm.attach(Lakebase("finance/filings"))
swarm.run_on(Compute.local("h100x8", burst="cloud:a100"))
swarm.guard(pii=True, budget_per_day=1200)

run = swarm.start(goal="Summarize Q3 risk factors across the portfolio")
for event in run.stream():
    print(event.agent, event.status, event.confidence)

$ curl -X POST https://api.stelux.ai/v1/swarms \
    -H "Authorization: Bearer $STELUX_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "name": "research-desk",
      "agents": [
        { "role": "planner" },
        { "role": "analyst", "replicas": 4 },
        { "role": "critic", "verifies": "analyst" }
      ],
      "data": ["lakebase://finance/filings"],
      "compute": { "local": "h100x8", "burst": "cloud:a100" },
      "guardrails": { "pii": true, "budget_per_day": 1200 }
    }'