Every idea goes through 5 phases of AI-powered research before we decide to build. Our research engine uses 9 specialized AI agents running in parallel, each performing 20+ web searches independently. This systematic approach has processed 425 ideas across 3 attack strategies.

Deep Research Methodology

Every idea goes through 5 phases of AI-powered research before we decide to build

425 Ideas Researched
600+ Avg Searches Per Idea
43K+ Avg Words Per Idea
9 AI Agents Per Phase
Research Pipeline
Each idea flows through five sequential phases — from raw data to final verdict
425 Niche Ideas
▼
Phase 0
📡
Data Enrichment
Pull Google Trends, keyword volumes, CPC benchmarks, and competitive signals for each idea
▼
Phase 1
🔍
Wide Scan
9 specialized AI agents launch in parallel — each performs 20+ web searches independently
Round 1 — Cast a Wide Net
9 agents run in parallel. Each explores its topic broadly with 20+ fresh Google searches.
🗣️
User Voices
20+ searches
🏿
Product Landscape
20+ searches
🔑
Keywords
20+ searches
📢
Ad Strategy
20+ searches
🔧
Build & Tech
20+ searches
💰
Monetiz.
20+ searches
⚠️
Stack Pitfalls
20+ searches
📊
Financial Intel
20+ searches
🏢
Competitive Intel
20+ searches
9 Reports Collected
Round 2 — Go Deep on Gaps
Each agent sees its own Round 1 report, then runs 15+ new searches to fill gaps. Zero repetition — every word must be net-new.
🗣️
User Voices
20+ searches
🏿
Product Landscape
20+ searches
🔑
Keywords
20+ searches
📢
Ad Strategy
20+ searches
🔧
Build & Tech
20+ searches
💰
Monetiz.
20+ searches
⚠️
Stack Pitfalls
20+ searches
📊
Financial Intel
20+ searches
🏢
Competitive Intel
20+ searches
▼
18 Independent Reports — 360+ Searches Total
▼
Phase 2
🔎
Gap Detection
1 auditor agent reads all 18 reports — finds contradictions, missing data, unanswered questions
▼
8–15 Follow-up Tasks Identified
▼
Phase 3
🎯
Deep Dives
8–15 targeted agents investigate each gap with 20+ focused searches
🔬
Deep Dive 1
🔬
Deep Dive 2
🔬
Deep Dive 3
···
🔬
Deep Dive N
▼
8–15 Targeted Reports — 300+ Additional Searches
▼
Phase 4
⚖️
Final Synthesis
Synthesizer scores idea 1–10 across weighted dimensions, produces GTM blueprint
BUILD
CONDITIONAL
KILL
Three Research Strategies
Each idea is researched through one of three strategic lenses, with agents tailored to that lens
🦈

Predator

Target incumbents where AI gives us a decisive cost or quality advantage. Find their weaknesses, exploit them.
101 ideas researched
  • Customer Complaints
  • Incumbent Analysis
  • Demand Validation
  • Switching Barriers
  • Clone Feasibility
  • Price Disruption
  • Stack Pitfalls
  • Financial Intelligence
  • Competitive Intel
💎

Underserved

Find problems with no good existing solutions. Serve users that incumbents ignore or underserve.
163 ideas researched
  • User Voices
  • Product Landscape
  • Keywords & Search Demand
  • Ad Strategy
  • Build & Tech
  • Monetization
  • Stack Pitfalls
  • Financial Intelligence
  • Competitive Intel
🌱

Provisioning

Spot emerging trends before solutions exist. Build the definitive tool before anyone else arrives.
161 ideas researched
  • Behavior Signals
  • Adjacent Solutions
  • Trend Demand
  • Market Entry
  • Build Feasibility
  • Market Economics
  • Stack Pitfalls
  • Financial Intelligence
  • Competitive Intel
Phase 1 Agents (Underserved Example)
Each agent is a specialist — here's what the 9 agents investigate for underserved niches
🗣️

User Voices

Scrapes Reddit, forums, and reviews for real user frustrations, workarounds, and language. Captures exact quotes for ad copy.

20+ searches × 2 rounds
🏿

Product Landscape

Maps every existing solution — apps, tools, templates. Analyzes pricing, review counts, ratings, and feature gaps.

20+ searches × 2 rounds
🔑

Keywords & Search Demand

Analyzes search volumes, CPCs, keyword difficulty, and long-tail opportunities. Estimates addressable search market.

20+ searches × 2 rounds
📢

Ad Strategy

Evaluates ad viability across Meta and Google. Estimates CAC, models creative angles, analyzes competitor ad patterns.

20+ searches × 2 rounds
🔧

Build & Tech

Assesses technical feasibility with our stack. Identifies API dependencies, AI model requirements, and potential blockers.

20+ searches × 2 rounds
💰

Monetization

Models pricing strategies, willingness-to-pay, LTV projections, and conversion benchmarks for the niche.

20+ searches × 2 rounds
⚠️

Stack Pitfalls

Checks for known technical gotchas — API rate limits, platform restrictions, legal constraints, data availability issues.

20+ searches × 2 rounds
📊

Financial Intelligence

Researches comparable product revenues via Google Search. Builds ARR estimates with CAC, LTV, and pricing models.

20+ searches × 2 rounds
🏢

Competitive Intel

Profiles competitor business health: team sizes, funding, ad spend, revenue estimates, AI-native readiness.

20+ searches × 2 rounds
ROAS-First Scoring Framework
Ideas are scored on weighted dimensions that prioritize ad economics and business sustainability
Ad Viability
20%
Can we profitably acquire users via Meta/Google ads?
Monetization
20%
Clear path to revenue — willingness to pay, LTV potential
Competition Gap
15%
Is the market open enough for a new AI-native entrant?
Competitive Intel
15%
How well-resourced and entrenched are existing competitors?
Build Feasibility
15%
Can we ship an MVP in 3-5 days with our stack?
Market Size
10%
Enough demand to sustain $10K+ MRR at scale?
Aha Speed
5%
How fast does the user feel magic after first interaction?
What Each Idea Produces
By the time an idea gets a verdict, we have a comprehensive research package
18
Phase 1 Reports
9 agents × 2 rounds
1
Gap Analysis
with 8-15 follow-up tasks
8-15
Deep Dive Reports
targeted investigations
1
Final Synthesis
verdict + GTM blueprint