Deep Research Mode: Multi-Hop AI Research Explained with FOMOA
January 14, 2026
8 min read
Tushar Agrawal
AI Research ToolAutomated ResearchDeep Research AIFOMOAMulti-Hop SearchResearch AutomationInformation Synthesis
FOMOA's deep research follows leads across 3 hops, detects conflicts between sources, and synthesizes comprehensive reports. Perfect for journalists and researchers.
What is Multi-Hop Research?
Traditional search gives you a list of links. You read them, find mentions of other sources, search for those, read more, and eventually piece together a complete picture.
Multi-hop research automates this entire process.
FOMOA's Deep Research mode doesn't just search once - it follows leads from initial results, extracts key entities, searches for more information about those entities, and synthesizes everything into a comprehensive answer.
How Multi-Hop Research Works
Traditional Search (Single Hop)
================================
Query: "Impact of UPI on Indian economy"
↓
Single search
↓
10 results
↓
Manual reading required
↓
User follows links manually
↓
Hours of work
FOMOA Deep Research (Multi-Hop)
================================
Query: "Impact of UPI on Indian economy"
↓
Query Expansion (5 related queries)
↓
Parallel Search (all 5 queries)
↓
Entity Extraction (NPCI, RBI, PhonePe...)
↓
Follow-up Search (deep dive on entities)
↓
Conflict Detection (flag contradictions)
↓
Synthesis (structured report)
↓
60 seconds total
Three Research Depths
FOMOA offers three research intensities:
1. Quick Mode (5 seconds)
Configuration:
- Queries: 2
- Hops: 1
- Sources: ~10
Best for:
- Factual lookups
- Quick verifications
- Simple questions
Example: "What is NPCI?"
2. Normal Mode (15 seconds)
Configuration:
- Queries: 4
- Hops: 2
- Sources: ~25
Best for:
- Background research
- Topic overviews
- Comparison queries
Example: "Compare UPI with international payment systems"
3. Deep Mode (60 seconds)
Configuration:
- Queries: 6
- Hops: 3
- Sources: ~50
Best for:
- Comprehensive research
- Investigative journalism
- Academic papers
- Complex policy analysis
Example: "Impact of UPI on Indian economy including
rural adoption, merchant digitization,
and international expansion plans"
The Research Process: Step by Step
Let's trace a real research query through FOMOA's system:
Query: "Impact of UPI on Indian economy 2024"
STEP 1: Query Expansion
=======================
Original: "Impact of UPI on Indian economy 2024"
Generated queries:
├── "UPI transaction volume growth 2024 statistics"
├── "UPI impact on cash transactions India RBI data"
├── "UPI merchant adoption rate small businesses"
├── "UPI international expansion NPCI plans"
├── "Digital payments GDP contribution India"
└── "UPI vs cash economy percentage 2024"
STEP 2: Parallel Search
=======================
All 6 queries searched simultaneously
Sources found:
├── rbi.org.in/digitalpaymentsstatistics
├── npci.org.in/statistics
├── pib.gov.in/PressRelease
├── economictimes.com/tech/UPI
├── livemint.com/fintech
├── moneycontrol.com/digital-payments
├── businesstoday.in/UPI
├── thehindu.com/business/Economy
├── worldbank.org/india/digitalpayments
└── imf.org/india/fintech
STEP 3: Entity Extraction
=========================
Key entities identified:
Organizations:
├── NPCI (National Payments Corporation of India)
├── RBI (Reserve Bank of India)
├── PhonePe, Google Pay, Paytm (UPI apps)
└── SEBI (market impact)
Metrics:
├── Transaction volume: 14 billion/month
├── Transaction value: ₹20 lakh crore/month
├── Active users: 350 million
└── Merchant QR codes: 320 million
Policies:
├── UPI Lite
├── UPI 123PAY (feature phones)
└── UPI Global (international)
STEP 4: Follow-up Search (Hop 2)
================================
Deep dive on extracted entities:
Search: "NPCI UPI international expansion 2024"
→ Found: Singapore, UAE, France, UK partnerships
Search: "UPI merchant digitization rural India"
→ Found: 67% growth in tier-3/4 cities
Search: "RBI digital payment statistics 2024"
→ Found: Official transaction data
Search: "UPI impact on cash economy"
→ Found: Cash-to-GDP ratio declining
STEP 5: Hop 3 (Deep Mode Only)
==============================
Third-level exploration:
From NPCI international data:
├── Search: "UPI Singapore launch statistics"
├── Search: "UPI UAE remittance impact"
└── Search: "NPCI France partnership details"
From rural digitization data:
├── Search: "PM SVANidhi UPI adoption"
├── Search: "Kirana store digital payment growth"
└── Search: "Rural internet penetration UPI"
STEP 6: Conflict Detection
==========================
Checking for contradictions:
Source A (News): "UPI transactions: 12 billion/month"
Source B (NPCI): "UPI transactions: 14.05 billion/month"
→ FLAG: Possible outdated data in Source A
Source C (Blog): "UPI market share: 85%"
Source D (RBI): "UPI market share: 67% by volume"
→ FLAG: Different metrics (volume vs value)
Resolution: Prioritize official sources (RBI, NPCI)
Note: Flag discrepancies in output
STEP 7: Synthesis
=================
Structured output generation:
{
"summary": "UPI has transformed India's payment landscape...",
"key_findings": [
{
"finding": "Transaction volume reached 14 billion/month",
"source": "npci.org.in",
"confidence": "high"
},
...
],
"statistics": {...},
"conflicting_data": [...],
"sources_used": [...],
"further_reading": [...]
}
API Usage
import requests
def deep_research(query: str, depth: str = "normal") -> dict:
"""
Perform multi-hop research with FOMOA
Args:
query: Research topic
depth: "quick" (5s), "normal" (15s), or "deep" (60s)
Returns:
Structured research report
"""
response = requests.post(
"https://fomoa.cloud/api/research",
json={
"query": query,
"depth": depth,
"include_sources": True,
"detect_conflicts": True,
"language": "auto" # Auto-detect Hindi/English
},
headers={"Authorization": "Bearer your_api_key"}
)
return response.json()
# Example usage
result = deep_research(
"Impact of UPI on Indian economy 2024",
depth="deep"
)
print(f"Summary: {result['summary']}")
print(f"Sources consulted: {len(result['sources'])}")
print(f"Conflicts detected: {len(result['conflicts'])}")
Response Structure
{
"query": "Impact of UPI on Indian economy 2024",
"depth": "deep",
"processing_time_seconds": 58.3,
"summary": "UPI has fundamentally transformed India's payment ecosystem, processing over 14 billion transactions monthly valued at ₹20+ lakh crore. Key impacts include...",
"key_findings": [
{
"finding": "Monthly UPI transactions exceeded 14 billion in December 2024",
"source": {
"url": "https://npci.org.in/statistics",
"credibility_score": 0.98
},
"confidence": "high"
},
{
"finding": "Rural UPI adoption grew 67% YoY in tier-3/4 cities",
"source": {
"url": "https://rbi.org.in/digitalreport",
"credibility_score": 1.0
},
"confidence": "high"
}
],
"statistics": {
"transaction_volume": "14.05 billion/month",
"transaction_value": "₹20.64 lakh crore/month",
"active_users": "350 million",
"merchant_qr_codes": "320 million",
"source": "NPCI December 2024"
},
"entities_found": [
{
"name": "NPCI",
"type": "organization",
"relevance_score": 0.95,
"brief": "National Payments Corporation of India, operates UPI"
},
{
"name": "UPI Lite",
"type": "product",
"relevance_score": 0.82,
"brief": "Offline-capable small-value UPI transactions"
}
],
"conflicts_detected": [
{
"topic": "UPI market share percentage",
"source_a": {
"claim": "85% of digital payments",
"url": "finance-blog.com",
"credibility": 0.55
},
"source_b": {
"claim": "67% by volume, 45% by value",
"url": "rbi.org.in",
"credibility": 1.0
},
"resolution": "RBI data more reliable; blog may use different metric"
}
],
"timeline": [
{
"date": "2016",
"event": "UPI launched by NPCI"
},
{
"date": "2024",
"event": "14 billion monthly transactions milestone"
}
],
"sources_used": [
{
"url": "https://npci.org.in/statistics",
"title": "NPCI UPI Statistics",
"credibility_score": 0.98,
"used_for": ["transaction_volume", "merchant_data"]
}
],
"further_reading": [
{
"topic": "UPI International Expansion",
"suggested_query": "UPI global NPCI international partnerships 2024"
}
],
"language_detected": "english",
"total_sources_analyzed": 47,
"hops_completed": 3
}
Real-World Use Cases
For Journalists
Query: "Adani group controversy timeline 2023-2024"
FOMOA Deep Research provides:
- Chronological timeline with dates
- Multiple source perspectives
- Stock price impact data
- Official company responses
- SEBI investigation updates
- Conflict flags where reports differ
For Researchers
Query: "Climate change impact on Indian agriculture"
Output includes:
- ICAR research citations
- Government policy responses
- Crop yield statistics by region
- Farmer adaptation strategies
- International comparison data
- Academic paper summaries
For Students
Query: "Indian Independence Movement key events"
Structured output:
- Timeline from 1857-1947
- Key figures with brief bios
- Important dates and events
- Multiple perspectives (Indian/British)
- Source citations for papers
For Policy Analysts
Query: "Comparison of healthcare schemes - Ayushman Bharat vs state schemes"
Comprehensive analysis:
- Coverage comparison table
- Beneficiary statistics
- Implementation challenges
- Budget allocations
- Success metrics by state
- Expert opinions
Comparison with Single-Search AI
Query: "EV policy India 2026"
Single Search (ChatGPT/Perplexity):
==================================
- Generic overview
- May miss recent updates
- No conflict detection
- Single perspective
- Limited source diversity
- 2-3 second response
FOMOA Deep Research:
====================
- Central + State policy breakdown
- FAME II subsidy details
- PLI scheme for batteries
- State-wise incentive comparison
- Industry response data
- Recent policy amendments
- Conflicting projections flagged
- 47 sources consulted
- 60 second response
Best Practices for Deep Research
1. Frame Specific Queries
Poor query: "Tell me about startups"
→ Too broad, unfocused results
Good query: "Indian AI startups funding trends 2024 Bangalore"
→ Specific, actionable research output
2. Use Appropriate Depth
Quick (5s): Factual questions
- "What is SEBI's role?"
- "When was RBI founded?"
Normal (15s): Topic overviews
- "Compare UPI and IMPS"
- "Overview of PLI schemes"
Deep (60s): Comprehensive research
- "Analysis of India's renewable energy policy"
- "Impact of GST on small businesses"
3. Review Conflicts
# Always check for conflicting information
result = deep_research("GST collection growth 2024")
if result['conflicts_detected']:
print("⚠️ Conflicting data found:")
for conflict in result['conflicts_detected']:
print(f" Topic: {conflict['topic']}")
print(f" Resolution: {conflict['resolution']}")
Rate Limits for Research API
Research Depth Rate Limit Typical Response Time
-------------- ---------- --------------------
Quick 60/minute 3-5 seconds
Normal 30/minute 12-18 seconds
Deep 10/minute 45-75 seconds
Note: Deep research consumes more resources
and has stricter rate limits
---
Transform hours of manual research into 60-second comprehensive reports.
Try FOMOA's Deep Research at fomoa.cloud.
Building research tools or need custom depth configurations? Connect on LinkedIn.