HoloRadarHoloRadar
Smart Pre-Search Intelligence

How It Works

Intent inference automatically routes your query to the best sources. No generic search — every query gets a customized research strategy.

Intent Inference Pipeline

Your query flows through 7 stages before you see results

1
User Topic
Example: "electric vehicles 2026"
2
_infer_intent()
Analyzes query structure, detects "vs" patterns, understands context
3
Smart Source Selection
Tech trends → HN + GitHub + Reddit | Products → Reddit + X + forums
4
Parallel Fetch (ThreadPoolExecutor)
Query all selected platforms simultaneously using native APIs
5
Weighted Reciprocal Rank Fusion
Combines results with source-specific weights (HN: 1.2x for tech, Reddit: 1.3x for products)
6
Reranker
Reorders by relevance, engagement signals, and recency
7
cluster_candidates(cluster_mode)
Groups similar findings into evidence clusters and themes

Smart Source Routing

Different query types get different source prioritization

Tech Trends

Query: "rust language adoption 2026"

Hacker NewsPriority 1
GitHubPriority 1
RedditPriority 2

Product Feedback

Query: "rayban meta smart glasses reviews"

RedditPriority 1
XPriority 1
BlogsPriority 2

Investment Research

Query: "anthropic claude 3 market sentiment"

GitHubPriority 1
XPriority 1
BlogsPriority 1

Breaking News

Query: "openai gpt-5 announcement"

XPriority 1
Hacker NewsPriority 1
BlogsPriority 2

Depth Modes

Control how comprehensive your research is

Quick Mode

Max 2 sources

Fast overview from the 2 most relevant platforms based on QUICK_SOURCE_PRIORITY. Ideal for quick checks and validation.

Default Mode

All credential-gated sources

Comprehensive search across all platforms with API access. Balanced breadth and processing time.

Deep Mode

Full subquery fanout

Maximum depth — queries every possible subquery variant across all platforms. Catches edge cases and long-tail discussions.

N-Pass Competitor Fanout

Special handling for comparison queries

Detects "vs" Patterns

When your query contains "vs" or "versus", the system runs one complete research pass per entity.

Query: "next.js vs remix 2026"
→ Pass 1: Research "next.js" across all sources
→ Pass 2: Research "remix" across all sources
→ Synthesize: Compare findings side-by-side
Result:Multi-angle comparison with independent evidence for each side, not just generic "X vs Y" articles.

Advanced Search Tags

Platform-specific filtering across all sources

Reddit

  • • Subreddit filtering
  • • Flair tags
  • • Upvote velocity
  • • Comment depth

X

  • • Hashtag tracking
  • • Engagement signals
  • • Reply sentiment
  • • Verified accounts

GitHub

  • • Language filters
  • • Topic tags
  • • Star velocity
  • • Issue activity

Hacker News

  • • Domain filters
  • • User reputation
  • • Comment quality
  • • Frontpage timing

Blogs/Substack

  • • Author filtering
  • • Publication patterns
  • • RSS feeds
  • • Subscriber count

Cross-Platform

  • • Hashtag tracking
  • • Entity mentions
  • • Sentiment alignment
  • • Temporal patterns

Why This Matters

Competitors use generic search: ChatGPT, Perplexity, and Gemini use the same web search approach for all queries. A tech trend gets the same source prioritization as product feedback.

HoloRadar uses smart routing: Intent inference means tech trends automatically prioritize Hacker News and GitHub, while product feedback prioritizes Reddit and X. You get better intelligence because the system understands what you are actually looking for.

Native APIs beat web scraping: Direct platform access means fresher data, complete metadata, and no crawl delays. When Google indexes a Reddit thread hours later, we have already analyzed it in real-time.

Ready to Try Smart Pre-Search Intelligence?

Join the waitlist for web app and iOS. Android available now.

Join Waitlist