Skip to main content
Practical tips for getting better results from the Valyu DeepSearch API.

​
Multi-Step Search Workflows

For complex research tasks, break your search into multiple steps rather than relying on a single query. This works especially well for technical domains like research, finance, and medicine. Example workflow: 
# Multi-step search workflow
async def research_agent(query: str):
    # Step 1: Break down the query into focused searches
    sub_queries = decompose_query(query)

    results = {}
    for i, sub_query in enumerate(sub_queries):
        # Step 2: Adjust strategy based on what you've found
        strategy = adapt_strategy(sub_query, results)

        search_result = valyu.search(
            query=sub_query,
            included_sources=strategy.sources,
            max_price=strategy.budget,
            relevance_threshold=0.65
        )
        results[f"step_{i}"] = search_result

        # Step 3: Fill in any gaps
        gaps = identify_knowledge_gaps(search_result, query)
        if gaps:
            gap_result = valyu.search(
                query=gaps[0].refined_query,
                included_sources=gaps[0].target_sources,
                max_price=50.0
            )
            results[f"gap_fill_{i}"] = gap_result

    # Step 4: Combine everything
    return synthesise_multi_source_findings(results)

​
AI vs Human Searches

Valyu is optimised for AI agents by default. The tool_call_mode parameter controls this: For AI agents (the default): 
# Optimised for LLMs
response = valyu.search(
    "quantum error correction surface codes LDPC performance benchmarks",
    tool_call_mode=True,  # Default
)
For human-facing searches: 
# Better for human readability
response = valyu.search(
    "quantum computing error correction methods",
    tool_call_mode=False,
)

​
Using Search Parameters

Combine good prompts with search parameters: 
response = valyu.search(
    "GPT-4 vs GPT-3 architectural innovations: training efficiency, inference optimisation, and benchmark comparisons",
    search_type="proprietary",
    max_num_results=10,
    relevance_threshold=0.6,
    included_sources=["valyu/valyu-arxiv"],
    max_price=50.0,
    category="machine learning",
    start_date="2024-01-01",
    end_date="2024-12-31"
)
Use included_sources to search datasets other APIs can’t access—like valyu/valyu-arxiv for academic papers or specialised datasets for financial data.

​
Balancing Quality and Cost

​
Budget Tiers

Not getting enough results? Try increasing max_price: 
search_configs = [
    {"max_price": 20.0, "use_case": "Quick fact-checking"},
    {"max_price": 50.0, "use_case": "Standard research"},
    {"max_price": 100.0, "use_case": "Comprehensive analysis"},
]
What each budget gets you:
  • $20 CPM: Basic web + academic content
  • $50 CPM: Full web + most research databases + financial data
  • $100 CPM: Premium sources + financial data + specialised datasets
Higher budgets unlock exclusive sources like academic journals, financial data feeds, and curated research databases.

​
Managing Context Size

Control how much data goes into your LLM’s context window: 
# Smaller context
lightweight_search = valyu.search(
    "transformer architecture innovations",
    max_num_results=3,
    results_length="short",
    max_price=50.0
)

# Larger context
comprehensive_search = valyu.search(
    "transformer architecture innovations",
    max_num_results=15,
    results_length="max",
    max_price=100.0
)
Token estimates:
  • Short: ~6k tokens per result (25k chars)
  • Medium: ~12k tokens per result (50k chars)
  • Long: ~24k tokens per result (100k chars)
  • Rule of thumb: 4 characters ≈ 1 token
Start with max_num_results=10 and results_length="short", then adjust based on your needs.

​
Specialised Datasets

Valyu offers datasets beyond standard web search. Browse them at platform.valyu.ai/data-sources. Categories include:
  • Academic: ArXiv, PubMed, academic publishers
  • Financial: SEC filings, earnings reports, market data
  • Medical: Clinical trials, FDA drug labels, medical literature
  • Technical: Patents, specifications, implementation guides
  • Books & Literature: Digitised texts, reference materials
Targeting specific datasets: 
# Academic search
academic_search = valyu.search(
    "CRISPR gene editing clinical trials safety outcomes",
    included_sources=["valyu/valyu-pubmed", "valyu/valyu-US-clinical-trials"],
    max_price=30.0
)

# Financial search
financial_search = valyu.search(
    "Tesla Q3 2024 earnings revenue breakdown",
    included_sources=["valyu/valyu-US-sec-filings", "valyu/valyu-US-earnings"],
    max_price=60.0
)
Data advantage: Proprietary datasets behind the DeepSearch API often contain information unavailable through standard web APIs, giving your AI system access to authoritative, structured knowledge that improves factual accuracy.

​
Common Mistakes to Avoid

  1. Wasting tokens: Use max_num_results and results_length to control context size
  2. Skipping filters: Set relevance thresholds and source controls
  3. Ignoring costs: Balance max_price with your quality needs
  4. Wrong sources: Pick datasets that match your domain—academic, financial, medical, or web
  5. Single-shot queries: For complex research, use multi-step workflows

​
Next Steps

Quick Integration

Get your first Valyu search running in minutes

API Reference

Explore all search parameters and response formats

​
Get Help