Deployment

An agent is deployed when you call .save(), and the same definition runs in the cloud or on-prem.

Mode	Where it runs	Use it for
Cloud	aixplain's managed cloud, serverless	The default. No infrastructure to manage, and scaling is handled for you.
On-prem	Your own infrastructure, including air-gapped networks	Data that must stay inside your network

Try it

Get a saved agent by its ID and run it. In production, read the key from an environment variable.

import os
from aixplain import Aixplain

aix = Aixplain(api_key=os.environ["AIXPLAIN_API_KEY"])

AGENT_ID = "YOUR_AGENT_ID"  # from agent.id after .save()

agent = aix.Agent.get(AGENT_ID)
result = agent.run("Ticket: The export button returns a 500 error since this morning.")
print(result.status)
print(result.data.output)

Show output

Call from any language

Agent runs are asynchronous over HTTP. Start the run, then poll the URL returned in data until completed is true.

import os
import time
import requests

AGENT_ID = "YOUR_AGENT_ID"  # from agent.id after .save()
headers = {"x-api-key": os.environ["AIXPLAIN_API_KEY"], "Content-Type": "application/json"}

start = requests.post(
    f"https://platform-api.aixplain.com/v2/agents/{AGENT_ID}/run",
    headers=headers,
    json={"query": "Ticket: The export button returns a 500 error since this morning."},
    timeout=30,
).json()

while True:
    result = requests.get(start["data"], headers=headers, timeout=30).json()
    if result["completed"]:
        break
    time.sleep(2)

print(result["status"])
print(result["data"]["output"])

Show output

The REST API page covers run parameters, multi-turn conversations and errors, with JavaScript and curl examples.

Cloud

Cloud is the default: every saved agent runs on aixplain's managed infrastructure and is billed per run. Watch usage, cost and run traces in the aixplain app.

Before you go live:

• Test the agent on representative and adversarial inputs, and read result.data.steps to confirm which tools ran.
• Add Inspectors for safety, quality and compliance checks.
• Cap cost, time and iterations per run with a budget.
• Scope access and quotas with API keys and Rate limiting.

On-prem

On-prem runs the aixplain platform inside your own infrastructure, including air-gapped networks with no outbound connectivity.

What you get	Details
Same agents	Agents, tools and governance work as they do in the cloud. Some services are not available on-prem.
Containerized services	Linux containers (Docker or Docker Compose), scaled behind a load balancer. You provision GPU nodes for model serving.
Air-gapped install	Signed offline bundles, signed patch bundles checked by SHA-256, and licenses that validate locally.
No staff access	aixplain staff have no access to your environment. Data, models, logs and telemetry stay in your network.

To scope an on-prem deployment, contact aixplain.

Next steps

• Triggers: run a deployed agent on a schedule or when an event fires.
• REST API: call agents and models from any language.
• API keys: create scoped keys for each environment.
• Rate limiting: cap requests, tokens and credits per key.