How to Add Real-Time News to GPT-4 Agents (2025 Guide)

Getting real-time news into GPT-4, Claude, or any OpenAI-compatible LLM used to mean building JSON function schemas by hand, scraping news sites, and maintaining brittle integrations. wpnews eliminates all of that with 69 pre-built tool schemas you can drop in with two lines of Python.

Why agents need real-time news

LLMs have a training cutoff. For anything happening after that date — new AI model releases, startup funding rounds, regulatory news, market moves — they either hallucinate or admit they don't know. The fix is retrieval-augmented generation (RAG) with live news.

But not all news APIs are built for LLMs. Most require you to:

• Write JSON function schemas manually (error-prone, maintenance burden)
• Parse and clean raw article HTML yourself
• Handle rate limits, caching, and error responses
• Build entity extraction and topic classification on top

wpnews does all of this for you and ships 67 ready-to-use OpenAI function schemas.

Setup: 2-line integration

First, get a free API key at wpnews.pro/api (no credit card, instant). Then:

import openai
import requests

# Load all 67 wpnews tool schemas — no manual JSON definition
tools = requests.get(
    "https://api.wpnews.pro/api/v1/openai-tools.json"
).json()

client = openai.OpenAI()
response = client.chat.completions.create(
    model="gpt-4o",
    tools=tools,
    messages=[{"role": "user", "content": "What's the biggest AI story today?"}],
)
print(response.choices[0].message.content)

That's it. GPT-4 will automatically call get_news, search_news, get_radar, or any of the other 69 tools based on what the user asks.

Handling tool calls (full agentic loop)

When GPT-4 decides to call a wpnews tool, you need to execute the call and return the result. wpnews provides a URL-resolver endpoint to make this easy:

import openai, requests, json

API_KEY = "YOUR_WPNEWS_KEY"
HEADERS = {"X-API-Key": API_KEY}
API_BASE = "https://api.wpnews.pro"

tools = requests.get(f"{API_BASE}/api/v1/openai-tools.json").json()
client = openai.OpenAI()

messages = [{"role": "user", "content": "Summarize AI funding news this week"}]

while True:
    response = client.chat.completions.create(
        model="gpt-4o", tools=tools, messages=messages
    )
    msg = response.choices[0].message

    if msg.tool_calls:
        messages.append(msg)
        for tc in msg.tool_calls:
            # Resolve tool name to API URL
            url_resp = requests.get(
                f"{API_BASE}/api/v1/openai-tools/{tc.function.name}/url",
                params=json.loads(tc.function.arguments),
                headers=HEADERS,
            ).json()
            # Fetch the actual news data
            news = requests.get(url_resp["url"], headers=HEADERS).json()
            messages.append({
                "role": "tool",
                "tool_call_id": tc.id,
                "content": json.dumps(news),
            })
    else:
        print(msg.content)
        break

Key tools for AI agent use cases

• get_news — latest articles filtered by topic, sentiment, and language
• search_news — full-text BM25 search with date range filtering
• get_radar — AI ecosystem temperature: trending topics, sentiment, top entity
• get_adaptive_context — auto-selects how much context to load (50-400 tokens) based on news activity
• get_news_velocity — detect news bursts before loading heavy context
• get_trending_entities — companies and people surging in coverage right now
• get_entity_brief — one-sentence snapshot of any named entity for system prompt injection
• get_watchlist_digest — monitor multiple entities/topics in a single call

Pattern: smart context loading

Instead of blindly loading news into every agent call, use get_news_velocity as a gate:

import requests

API_KEY = "YOUR_WPNEWS_KEY"
HEADERS = {"X-API-Key": API_KEY}

velocity = requests.get(
    "https://api.wpnews.pro/api/v1/news-velocity",
    headers=HEADERS
).json()

if velocity["is_burst"]:
    # Breaking news — load full context (400 tokens)
    ctx = requests.get(
        "https://api.wpnews.pro/api/v1/adaptive-context?force_mode=surge",
        headers=HEADERS
    ).json()
elif velocity["is_quiet"]:
    # Slow news day — minimal context (50 tokens)
    ctx = requests.get(
        "https://api.wpnews.pro/api/v1/adaptive-context?force_mode=quiet",
        headers=HEADERS
    ).json()
else:
    # Normal — auto-budget
    ctx = requests.get(
        "https://api.wpnews.pro/api/v1/adaptive-context",
        headers=HEADERS
    ).json()

system_prompt = f"Current AI news context: {ctx['context_text']}"

Pricing and limits

• Free: 1,000 requests/day — enough for most development and low-traffic production
• Pro ($19/mo): 10,000 req/day, webhooks, 50-item watchlist, Pro-only endpoints

Try the demo endpoint without any key: curl https://api.wpnews.pro/api/v1/demo