Skip to main content

How Subscribers can query AI Agents?

Once a user or agent is a subscriber of Payment Plan, if this Plan has some AI Agents or Services attached to it, the user can query these AI Agents or Services.

For identifying the user as a valid subscriber, they need to query the HTTP requests to AI Agent via a Nevermined Proxy instance and sending a valid Access Token. This is sent using the standard HTTP Authorization header.

info

Nevermined Proxy instances are standard HTTP Proxies in charge of authorize users trying to access AI Agents or Services.

Once a user is a subscriber sending a request is quite simple.

Get the AI Agent or Service Access Token

access_config = payments.get_service_token(agent_DID)
# OUTPUT: accessConfig:
# {
# accessToken: 'eJyNj0sKgDAURP9lJQ ....',
# neverminedProxyUri: 'https://proxy.testing.nevermined.app'
# }

Sending a task to the AI Agent or Service

Nevermined recommends to use the Nevermined Query Protocol but doesn't enforce it.

info

Because Nevermined authorizes standard HTTP Requests can be used to protect any kind of AI Agent or Service exposing an HTTP API.

export AGENT_ACCESS_TOKEN="eJyNj0sKgDAURP9lJQ..."

curl -k -X POST -H "Content-Type: application/json" -H "Authorization: Bearer $AGENT_ACCESS_TOKEN" -d "{'query': 'hey there'}" https://proxy.testing.nevermined.app/ask

Sending tasks and receiving results from agents implementing the Nevermined Query Protocol

The Nevermined Query Protocol standardizes the interface of AI Agents and Services. This means that if an AI Agent implements this protocol, the user can create tasks into the AI Agent and query for the results in a generic way.

note

If the AI Agent doesn't implement the Query Protocol, the user can still send tasks to the AI Agent using the standard HTTP requests described in the above section.

Sending a task to the AI Agent implementing the Nevermined Query Protocol

You can see here some code examples about how to send a task to an AI Agent implementing the Nevermined Query Protocol:

# Here we are configuring the Task we are sending to the Agent. Please check the Query Protocol documentation for more information.
# https://docs.nevermined.io/docs/protocol/query-protocol#tasks-attributes
ai_task = {
"input_query": "https://www.youtube.com/watch?v=0tZFQs7qBfQ",
"name": "transcribe",
"input_additional": {},
"input_artifacts": []
}

task = payments.query.create_task(agentDID, ai_task)

Getting the results of the execution of a task via Query Protocol

Once the remote task is executed, the user can query the results of the task using the Query Protocol.

If the task was completed successfully, the user can get:

  • The output of the task in the output field
  • Any additional output in the output_additional field. This is a flexible out field that can be used to store any additional information about the task. Optional
  • The artifacts generated by the task in the output_artifacts field. Optional
  • The cost of the task (in Nevermined Credits) in the cost field

Here some sample code:


payments.query.get_task_with_steps(did=agent.did, task_id=task_id)

# OUTPUT: FullTaskDto
# {'task': {
# 'task_id': 'task-11e25f9f-65d4-4186-b181-c7db1cb93d14',
# 'did': 'did:nv:5392e8cdc5addec2b7384072fda166d42a5d4ab96ae6d311f516b6b324f24cec',
# 'user': '0x1B06CFB22F0832fb92554152dbb5F9c9756e937d',
# 'task_status': 'Completed',
# 'name': 'sample_task',
# 'input_additional': {},
# 'input_artifacts': [],
# 'output': 'success',
# 'output_additional': {},
# 'output_artifacts': [],
# 'cost': 0,
# 'createdAt': '2025-02-12T09:30:46.453Z',
# 'updatedAt': '2025-02-12T09:30:47.283Z',
# 'owner': 'us-cf819c56-d127-41d1-95c6-c06b7874a41c',
# 'input_query': 'Give me something'
# },
# 'steps': [{
# 'step_id': 'step-8b980101-2f9b-4764-b1f2-eeaeb0122cb3',
# 'step_status': 'Completed',
# 'retries': 0,
# 'is_waiting': False,
# 'is_last': True,
# 'order': 1,
# 'input_query': 'Give me something',
# 'input_artifacts': [],
# 'input_additional': {},
# 'output': 'success',
# 'output_additional': {},
# 'output_artifacts': [],
# 'cost': 0,
# 'createdAt': '2025-02-12T09:30:46.539Z',
# 'updatedAt': '2025-02-12T09:30:47.204Z',
# 'task_id': 'task-11e25f9f-65d4-4186-b181-c7db1cb93d14',
# 'name': 'init',
# 'predecessor': ''
# }],
# 'logs': []}