First run — Instant proof, then a job

First run (do this)

• Open Today → Instant proof (~1 min). No canvas, no model picker required.
• Open Work OS and describe a real job. Approve the plan.
• Open Proof for status, cost when available, and the trail.

Optional — connect an external agent

Skip this unless you already have an agent outside FuseIQ. Most users never need the SDK on day one.

1. Install the FuseIQ SDK

Open your terminal and install the SDK:

bash
pip install fuseiq-sdk

2. Get Your API Key

Navigate to Settings → API tokensin your dashboard. Generate a new key. Keys are prefixed fk_live_... — copy it immediately as it is shown only once.

3. Add the Agent Wrapper

Wrap any agent function with the FuseIQ heartbeat. Here's a minimal example using the current API:

python
from fuseiq import FuseIQAgent
import os

agent = FuseIQAgent(
    api_key=os.getenv("FUSEIQ_API_KEY"),
    agent_name="My First Agent",
    agent_type="llm",
    description="A helpful assistant for customer queries",
    heartbeat_url="https://fuseiq.io/api/external/heartbeat",
)

@agent.on_heartbeat
def report_status():
    return {
        "status": "online",
        "load": 0.45,
        "tasks_completed": 128,
        "uptime_seconds": 3600,
    }

@agent.on_message
def handle_request(payload):
    # Your agent logic here
    response = your_agent_fn(payload["message"])
    return {"response": response}

agent.start()

4. Verify in Staff Directory

Within 30 seconds, your agent will appear in the Staff Directorywith a green online indicator. You can now:

• Assign tasks directly from Ops Kanban
• Monitor cost and performance in Analytics
• Add the agent to a Swarm workflow

Supported Providers

FuseIQ supports the following LLM providers for agent execution:

• OpenAI (GPT-4o, GPT-4o-mini, o-series)
• Anthropic (Claude Sonnet 4, Opus)
• Google (Gemini Pro 2.5, Gemini Flash)
• Grok (xAI)
• Kimi (Moonshot AI)
• Perplexity (Sonar, Sonar Pro)
• Manus
• Custom / Self-hosted models