Integration Pattern

AI Agent Grounding

AI billing agents hallucinate rules, invent modifier logic, and confuse payment systems. RCIntel gives your agents a source-linked knowledge source over MCP or REST, so every answer traces back to a citation instead of stale training data.

For: AI RCM platforms, voice AI, coding assistants, chatbot builders

MCP is in preview. Live MCP tools today cover 340B policy; payment, coding, and payer data are available over REST now and arrive as MCP tools as they reach preview.

How it works

architecture flow

Your AI Agent (Claude, GPT, custom LLM)
  │
  ├─ User asks a billing question, or the agent needs context
  │
  ├─ Agent calls RCIntel over MCP or REST
  │   ├─ resolve_340b_restrictions   → 340B contract-pharmacy limits  (MCP)
  │   ├─ /v1/codes/{code}/payment    → source-linked reimbursement    (REST)
  │   ├─ /v1/codes/{code}/ncci       → PTP edits + MUE limits          (REST)
  │   └─ /v1/payers/{slug}/...       → filing rules, denial codes      (REST)
  │
  ├─ Agent receives structured, source-linked data
  │
  └─ Agent answers with grounded, cited facts

Connect via MCP Server

Add the RCIntel MCP server to your agent's configuration. Works with Claude, Cursor, and any MCP-compatible agent.

MCP client config

{
  "mcpServers": {
    "rcintel": {
      "type": "http",
      "url": "https://api.your-domain.com/v1/mcp",
      "headers": {
        "X-API-Key": "YOUR_KEY"
      }
    }
  }
}

Or call the REST API directly

If your agent framework doesn't support MCP, wrap the same paths as tools over HTTP.

python

import httpx

class RCIntelTool:
    """Give an LLM agent access to source-linked billing facts."""

    def __init__(self, api_key: str):
        self.client = httpx.AsyncClient(
            base_url="https://api.your-domain.com",
            headers={"X-API-Key": api_key},
        )

    async def payment(self, code: str, ccn: str, payer: str = "medicare"):
        """Source-linked reimbursement for a code at a facility."""
        resp = await self.client.get(
            f"/v1/codes/{code}/payment",
            params={"ccn": ccn, "payer": payer},
        )
        return resp.json()

    async def coding_edits(self, code: str):
        """PTP edits and MUE limits for a code."""
        resp = await self.client.get(f"/v1/codes/{code}/ncci")
        return resp.json()

What your agent can call

resolve_340b_restrictions

MCP · live

340B manufacturer contract-pharmacy restrictions, source-linked.

list_340b_changes

MCP · live

Recent 340B policy changes with effective dates and citations.

/v1/codes/{code}/payment

REST · available

Source-linked Medicare/Medi-Cal reimbursement by facility, payer, and code.

/v1/codes/{code}/ncci

REST · preview

PTP edits and MUE limits for a code pair.

/v1/payers/{slug}/filing-deadline

REST · preview

Payer timely-filing and appeal windows.

/v1/npi/{npi}

REST · preview

Provider and organization records from the public registry.

Why agents need live knowledge

Training data

• Fee schedules go stale (updated annually)
• Payer rules change quarterly
• NCCI edits update every quarter
• No facility-specific context
• Hallucinated modifier requirements

RCIntel data layer

• CMS data updated as it changes
• Payer rules tracked as they change
• NCCI edits current to the quarter
• Facility-specific via CCN
• Structured, source-linked responses

Ground your agents in real data

Connect to the MCP server or call the REST API directly. Start with a free read-only key.

Get an API key Talk to us