How to Give a LangChain Agent B2B Data (With LeadOcean)

export LEADOCEAN_API_KEY="paste-your-key-here"
pip install langchain langchain-anthropic requests

"""leadocean_agent.py"""
import os, time, requests

BASE = "https://api.leadocean.io"
HEADERS = {"x-api-key": os.environ["LEADOCEAN_API_KEY"]}
BUDGET = 25          # stop the run after this many records
spent = 0

def call(method, path, body=None, params=None, cost=0):
    global spent
    if spent + cost > BUDGET:
        return {"error": "budget_reached", "spent": spent, "budget": BUDGET}
    for attempt in range(3):
        r = requests.request(method, BASE + path, headers=HEADERS,
                             json=body, params=params, timeout=30)
        wait = int(r.headers.get("Retry-After", "1"))
        if r.status_code == 429 and wait <= 2:
            time.sleep(wait)                  # over the per-second rate
            continue
        if r.status_code == 503:
            time.sleep(2 ** attempt)          # transient, never billed
            continue
        break
    try:
        data = r.json()
    except ValueError:
        return {"error": r.status_code, "message": "non-JSON response"}
    if r.status_code == 402:
        return {"error": "records_spent",
                "message": "Free records are used up. Stop and tell the user to upgrade."}
    if r.status_code == 429:
        return {"error": "paced",
                "message": f"Retry-After {wait}s. Stop and tell the user."}
    if not r.ok:
        return {"error": r.status_code,
                "message": data.get("error", {}).get("message")}
    spent += data.get("meta", {}).get("credits", 0)
    return data

from langchain.tools import tool

@tool
def count_people(job_level: str, job_function: str, country: str) -> dict:
    """Count people for a job level, job function and ISO country code.
    Free. Always call this before search_people."""
    body = {"jobLevel": [job_level], "jobFunction": [job_function],
            "country": [country], "limit": 1}
    res = call("POST", "/v1/people/search", body, params={"count": "true"})
    return res.get("meta", res)

@tool
def search_people(job_level: str, job_function: str, country: str, limit: int = 5) -> list:
    """Search people. Costs one record per person returned. Max 10 per call."""
    limit = max(1, min(limit, 10))
    body = {"jobLevel": [job_level], "jobFunction": [job_function],
            "country": [country], "limit": limit}
    res = call("POST", "/v1/people/search", body, cost=limit)
    if "error" in res:
        return res
    return [{"name": p["profile_data"].get("profile_full_name"),
             "headline": p["profile_data"].get("profile_headline"),
             "email_status": p["contact_data"].get("email_status"),
             "person_id": p["meta"]["source_ids"]["person_id"]}
            for p in res["data"]]

@tool
def enrich_person(person_id: str) -> dict:
    """Get emails and phones for one person_id from search_people. Costs 1 record.
    A person with nothing on file is a 404 and costs nothing."""
    res = call("POST", "/v1/people/enrich",
               {"person_id": person_id, "reveal_email": True}, cost=1)
    if "error" in res:
        return res
    return res["data"]["contact_data"]

from langchain.agents import create_agent

agent = create_agent(
    model="claude-sonnet-5",
    tools=[count_people, search_people, enrich_person],
    system_prompt=(
        "You find B2B contacts with LeadOcean tools. "
        "Call count_people first and report the total. "
        "Search at most 5 people, then enrich at most 2 who look best. "
        "If a tool returns an error, stop and explain it. Never retry a 402 or a pace."
    ),
)

if __name__ == "__main__":
    result = agent.invoke({"messages": [{"role": "user", "content":
        "Find VP-level sales leaders in Canada. Size the audience, "
        "show 5, and get contact details for the 2 best."}]})
    print(result["messages"][-1].content)
    print(f"Records spent this run: {spent} of {BUDGET}")

python leadocean_agent.py

pip install "langchain[mcp]"

from langchain.agents import create_agent
from langchain.mcp import MCPAdapter

async def main():
    async with MCPAdapter("https://api.leadocean.io/mcp") as adapter:
        tools = await adapter.list_tools()
        agent = create_agent("claude-sonnet-5", tools)
        return await agent.ainvoke({"messages": [{"role": "user",
            "content": "Count VP sales leaders in CA. Do not search."}]})

{"name":"job_level","total":6,"matched":6,"values":["C-Team","Director","Manager","Other","Staff","VP"],"truncated":false}

{
  "success": true,
  "data": [{
    "profile_data": {"profile_full_name": "Jane Doe", "profile_headline": "VP Sales at Acme"},
    "contact_data": {"has_email": true, "email_status": "verified", "email_type": "work"},
    "meta": {"source_ids": {"person_id": "1234567890"}}
  }],
  "meta": {"count": 1, "credits": 1, "nextCursor": "opaque-string"}
}