Research Principles
These principles apply to ALL research agents. Every agent must follow them.
Iterative Deep Research
Each agent performs 5-8 web searches minimum, organized in sequential rounds that drill deeper:
- Round 1: Broad overview queries
- Round 2: Drill-down into specific findings from Round 1
- Round 3: Cross-reference and validate
- Round 4: Reality check and edge cases
Do NOT stop after a single query. The first search gives you the surface — the follow-ups give you the insight.
Source Quality Tiers
Rank every finding by source reliability:
| Tier | Source Type | Use For |
|---|---|---|
| Tier 1 | Industry reports (Gartner, Forrester, McKinsey, IBISWorld), SEC filings, government data, Crunchbase verified data | Hard numbers, market share, verified metrics |
| Tier 2 | Reputable tech press (TechCrunch, Bloomberg, WSJ), company press releases, investor presentations, PitchBook | Funding data, company news, expert opinions |
| Tier 3 | Blog posts, Reddit threads, individual reviews, social media, founder interviews | Sentiment, narrative hooks, anecdotal evidence |
For pitch research, Tier 2 and Tier 3 sources are especially valuable — investor blog posts, demo day recaps, and founder stories reveal what narratives resonate and what language investors respond to.
Pitch-Specific Search Strategies
When researching investor preferences:
- Search for "{investor/fund name} portfolio", "{fund} thesis", "{investor} blog pitch advice"
- Look for demo day recaps and pitch competition results in the relevant space
When researching comparable narratives:
- Search for "{competitor} pitch deck", "{similar company} fundraise", "{space} demo day"
- Look for "X for Y" analogies that have worked for similar companies
Cross-Referencing
Never trust a single source for important claims. For every key finding:
- Look for 2-3 independent sources
- If sources agree: note convergence and cite all
- If sources disagree: note both, explain the discrepancy, and state which you trust more and why
Quantification
Vague claims are worthless — and investors see through them instantly:
- Bad: "They raised a big round"
- Good: "$15M Series A led by a16z in March 2025, $3M seed in 2023"
- Bad: "The market is growing"
- Good: "$4.2B in 2025, projected to reach $8.1B by 2028 at 12.3% CAGR (source: Grand View Research)"
Dating
Always note when data was published. Flag anything older than 12 months as potentially outdated. Fundraising landscapes shift fast — a market correction or new competitor can change everything in weeks.
Handling Research Failures
Sometimes WebSearch won't find what you need:
- Try alternative queries. Rephrase, use synonyms, try different angles. At least 3 variations before declaring a gap.
- Use proxy data. If you can't find a company's exact metrics, estimate from team size, funding, pricing × estimated customers. Show your math.
- Declare the gap explicitly. Write: "DATA GAP: Could not find reliable data on [X]. Closest proxy: [Y]. Confidence: Low."
- Never fabricate. An honest "unknown" is infinitely more valuable than a made-up number.
- Suggest how to fill the gap. Recommend specific research the founder can do to fill the gap before the pitch.