Skip to main content

​
Overview

Real-Time News brings current market developments into Finance Agent using Tavily search. When you need information about very recent events (last few days/weeks) that haven’t appeared in earnings calls or SEC filings yet, this feature searches live web sources and provides AI-generated summaries with article citations.

Live Web Search

Searches current web sources for breaking news and recent developments

AI Summaries

Tavily generates concise summaries of search results automatically

Source Citations

Every news item includes [N1], [N2], [N3] markers linking to original articles

Automatic Routing

Finance Agent detects news-related queries and invokes Tavily automatically

​
When to Use Real-Time News

Use this feature when you need information about:
  • “What’s the latest news on $NVDA?”
  • “Recent developments at $TSLA”
  • “Breaking news about $AAPL iPhone sales”
  • “How did the market react to Microsoft’s AI announcement?”
  • “Stock price movement after earnings miss”
  • “Investor sentiment on recent guidance cut”
  • “Recent acquisition announcements from $GOOGL”
  • “Partnership news for $CRM”
  • “M&A activity in the semiconductor sector”
  • “CEO transition news at $DIS”
  • “Recent executive departures”
  • “New board member announcements”
  • “Latest product announcements from $AAPL”
  • “New AI features from $MSFT”
  • “Recent service launches”

​
How It Works

Real-Time News is automatically triggered when:
  1. Your query contains news keywords (“latest news”, “recent”, “breaking”)
  2. The agent determines during iteration that news context would improve the answer
When activated:
  1. Query Tavily API - Sends your question to Tavily’s web search engine
  2. Retrieve Articles - Gets top 5 most relevant news articles with:
    • Title
    • URL
    • Content excerpt
    • Published date
  3. Generate Summary - Tavily’s AI creates a concise summary of findings
  4. Format with Citations - Articles are numbered [N1], [N2], [N3] for easy reference
  5. Merge into Answer - News context is incorporated into the final response

​
Query Examples

​
Latest News Lookup

Question: What's the latest news on $NVDA?

Answer:
## Latest NVIDIA News

**AI Chip Demand Remains Strong**
NVIDIA's Blackwell GPU platform is seeing unprecedented demand, with shipments 
expected to begin in Q4 2024 [N1]. CEO Jensen Huang stated that production is 
ramping ahead of schedule [N1].

**Partnership with Major Cloud Providers**
NVIDIA announced expanded partnerships with AWS, Azure, and Google Cloud to deliver 
Blackwell-powered AI infrastructure [N2]. The deals are valued at over $10B [N2].

**Stock Performance**
NVIDIA shares rose 7% this week on strong data center demand [N3]. Analysts cite 
continued AI adoption as a key driver [N3].

---

**Sources:**
- [N1] "NVIDIA's Blackwell GPUs Ship Ahead of Schedule" - Reuters, Dec 18 2024
- [N2] "Cloud Giants Expand NVIDIA Partnership" - Bloomberg, Dec 17 2024
- [N3] "NVIDIA Shares Rally on AI Optimism" - CNBC, Dec 16 2024

​
Combining News with Transcript Data

Question: How is $DDOG performing recently?

Answer:
## Datadog Recent Performance

**Q3 2024 Earnings (from transcript)**
Datadog reported **$748M** in billings for Q3 2024, up **21%** YoY [TC-1]. Revenue 
was **$690M**, beating guidance [TC-2]. Management raised full-year guidance to 
**$2.68B** in revenue [TC-3].

**Recent News Developments**
Datadog announced a new AI observability platform on Dec 15, 2024 [N1]. The product 
targets enterprise AI workloads and integrates with major LLM providers [N1].

Analysts raised price targets following the product launch, citing expansion into 
the growing AI infrastructure market [N2]. Stock is up 12% in December [N2].

**Sources:**
- [TC-1][TC-2][TC-3] Datadog Q3 2024 Earnings Call Transcript
- [N1] "Datadog Launches AI Observability Platform" - TechCrunch, Dec 15 2024
- [N2] "Analysts Boost Datadog Targets on AI Push" - Barron's, Dec 16 2024

​
Features

​
Automatic Semantic Routing

Finance Agent’s question analyzer detects when news data would be valuable: Explicit News Queries (always triggers news):
  • “latest news”
  • “recent developments”
  • “breaking news”
  • “what’s happening with”
Implicit News Needs (triggered during iteration):
  • Agent determines answer is incomplete without recent context
  • Question asks about very recent events (“this week”, “today”)
  • No relevant transcript/10-K data found

​
Hybrid Mode

Finance Agent can combine news with other data sources: News + Transcripts:
  • Start with earnings call data for fundamentals
  • Add news context for recent developments post-earnings
News + 10-K:
  • Use 10-K for historical financial position
  • Add news for recent strategic changes
News Only:
  • Pure news queries focus exclusively on Tavily results

​
Citation Format

News articles use [N1], [N2], [N3] markers:
  • [N1] = First news article retrieved
  • [N2] = Second news article
  • Citations include title, source, published date, and URL

​
AI-Generated Summaries

Tavily provides two types of output:
  • Answer: AI-generated summary of all findings (concise overview)
  • Results: Individual articles with titles, URLs, excerpts, and dates
Finance Agent incorporates both into the final response.

​
Coverage

  • Sources: Web-wide news search (financial news sites, press releases, company announcements, market analysis)
  • Freshness: Real-time (typically minutes to hours old)
  • Results: Top 5 most relevant articles per query
  • Provider: Tavily API

​
Technical Details

​
Tavily Integration

class TavilyService:
    def search_news(query: str, max_results: int = 5):
        # Query Tavily API
        response = tavily_client.search(
            query=query,
            max_results=max_results,
            include_answer="advanced"
        )
        
        return {
            "answer": "AI-generated summary",
            "results": [
                {
                    "title": "Article headline",
                    "url": "https://...",
                    "content": "Article excerpt",
                    "published_date": "2024-12-18",
                    "score": 0.95
                }
            ]
        }

​
News Context Format

News is formatted with the same SOURCE pattern as transcripts/10-K: 
=== LATEST NEWS ===

Summary: [Tavily AI-generated summary]

SOURCE [N1] Article Title (Published: Dec 18 2024, Source: https://...)
Article content excerpt (up to 800 characters)...

SOURCE [N2] Article Title (Published: Dec 17 2024, Source: https://...)
Article content excerpt...

=== END NEWS ===
This consistent format ensures the LLM cites news the same way as other sources.

​
Best Practices

1

Use News for Recent Events Only

Don’t rely on news for historical data. Use earnings transcripts or 10-K filings for past quarters/years.
2

Combine with Official Sources

For the best analysis, pair news with official data:
  • “Compare $AAPL’s Q3 2024 results with recent iPhone news”
3

Verify via Citations

Always check the [N1], [N2] citations to verify sources and read full articles.
4

Be Specific About Recency

Include time references:
  • “News from this week” (more targeted)
  • “Latest news” (broader)

​
Limitations

  • No Historical Archive: Tavily focuses on recent/current news, not historical events
  • Web Content Variability: Quality depends on available web sources at query time
  • No Structured Data: News is unstructured text, not official financial filings
  • Source Reliability: Tavily aggregates from multiple sources; verify important claims
  • No Real-Time Prices: For stock prices, use a dedicated market data service
Important: News search does NOT provide real-time stock prices or market data. It searches for news articles and press releases only.

​
Configuration

Real-Time News requires a Tavily API key: 
TAVILY_API_KEY=your_tavily_api_key_here
If the key is not configured:
  • News search will be disabled
  • Queries will fall back to transcripts/10-K only
  • A warning will appear in logs

Build docs developers (and LLMs) love

Get started for freeTalk to us