Telepath AI
MCP Orchestrator

Proactive NOC — Alarm to Ticket in Seconds

A fiber cut occurs on the Springfield OLT. The AI agent detects the alarm, identifies affected customers, cross-references their SLA tiers, auto-creates a trouble ticket in your ITSM platform (ServiceNow or Jira), and prioritises the response - all in a single conversation spanning ENM (network), BSS (customer) and ITSM systems.

Try These Prompts:

Expected Results:

Zero-Touch Service Activation

A new customer wants fiber service. The AI qualifies the address, shows available products with pricing, creates the order, activates the service in the network, and confirms the customer equipment is online — end-to-end orchestration via natural language.

Try These Prompts:

Expected Results:

Pre-Sales Service Qualification

A sales rep needs to quickly qualify an enterprise opportunity. Instead of waiting days for a response, the AI agent checks network topology, verifies capacity, qualifies the service, and retrieves eligible products — all in one conversation.

The agent validates the address, checks nearest network elements, verifies capacity on local aggregation nodes, performs a formal service qualification (TMF637), and retrieves eligible products from the catalog (TMF620).

Network Change Impact Prediction

An engineer is planning maintenance on a cell tower tonight and needs to understand the blast radius before taking it down.

The agent identifies the eNodeB, checks connected cells, handover configurations, assesses neighbour capacity, lists dependent services and customers, and checks pending orders — then recommends an optimal maintenance window.

Trouble Ticket Lifecycle - ServiceNow or Jira

Manage the full incident lifecycle through natural language. The AI agent creates, enriches, updates and resolves trouble tickets in a live ServiceNow or Jira Cloud instance, automatically correlating network faults and customer data from the BSS and ENM systems. The Jira integration also includes Confluence knowledge base search.

Try These Prompts:

Expected Results:

Sovereign GPU Infrastructure Provisioning

A customer needs an H100 GPU cluster for AI training in a specific sovereign data centre. The AI agent checks GPU availability across data centres, provisions the server, allocates GPUs, configures networking and storage - and links everything back to the customer's BSS account for billing.

Try These Prompts:

Expected Results:

OpenAI — Agents SDK

OpenAI added MCP support via their Agents SDK. Use the SSE transport to connect:

from agents import Agent
from agents.mcp import MCPServerSse

bss_server = MCPServerSse(
    url="https://akuig.bss.ngrok.dev/mcp"
)
enm_server = MCPServerSse(
    url="https://akuig.ngrok.io/mcp"
)
itsm_servicenow = MCPServerSse(
    url="https://akuigservicenow.ngrok.dev/mcp"
)
itsm_jira = MCPServerSse(
    url="https://akuigjiramcp.ngrok.dev/mcp"
)
infra_server = MCPServerSse(
    url="https://akuig.baremetal.ngrok.dev/mcp"
)

agent = Agent(
    name="Telepath NOC Agent",
    instructions="You are a telecom operations assistant...",
    mcp_servers=[bss_server, enm_server, itsm_servicenow, itsm_jira, infra_server]
)

Mistral — Le Chat & Agents API

Mistral supports MCP connectors in Le Chat and via their agents API:

from mistralai import Mistral

client = Mistral(api_key="your-key")

response = client.agents.complete(
    agent_id="your-agent-id",
    messages=[{"role": "user", "content": "Check network alarms"}],
    tools=[
        {"type": "mcp", "server_url": "https://akuig.bss.ngrok.dev/mcp"},
        {"type": "mcp", "server_url": "https://akuig.ngrok.io/mcp"},
        {"type": "mcp", "server_url": "https://akuigservicenow.ngrok.dev/mcp"},
        {"type": "mcp", "server_url": "https://akuigjiramcp.ngrok.dev/mcp"},
        {"type": "mcp", "server_url": "https://akuig.baremetal.ngrok.dev/mcp"}
    ]
)