How Our AI System Works

Radical transparency into how we analyze companies and generate investment assessments.

Our AI Doesn't Predict — It Synthesizes

We combine multiple data sources and AI agents to help you understand a company's situation. The goal is to surface insights you might miss, not to tell you what will happen.

Where We Get Our Data

Our system ingests data from multiple sources to build a complete picture of each company.

Company Reports

Quarterly and annual filings, earnings releases, and investor presentations.

News & Events

Press releases, regulatory announcements, and market news from trusted sources.

Market Data

Real-time prices, trading volumes, analyst estimates, and technical indicators.

Macro Indicators

Interest rates, commodity prices, currency movements, and economic indicators.

The Two-Layer Architecture

Our AI works in two stages: first, specialized "domain experts" analyze specific data types. Then, an "investment team" synthesizes everything into a coherent assessment.

Data Sources
Reports
News
Market
Macro
Layer 1: Domain Synthesis
Qualitative
Quantitative
News Synthesis
Macro Synthesis
Weekly Intel
Sector Analysis
Layer 2: Investment Committee
Sentiment
Risk/Opportunity
Data Quality
Reconciliation
Charting
Orchestrator
Final Investment Assessment

Layer 1: The Domain Experts

Think of Layer 1 as organizing all the puzzle pieces before solving the puzzle. Each agent focuses on a specific type of analysis.

AgentWhat It AnalyzesWhen It Runs
Qualitative
Business model, competitive position, management quality, strategic risksWhen a new report is downloaded
Quantitative
Financial statements, KPIs, peer comparisons, valuation metricsWhen new period data is available
News Synthesis
Daily market news and 4-day aggregated trendsDaily and every 4 days
Macro Synthesis
Interest rates, commodities, currency, economic indicatorsWeekly
Weekly Intel
Rolling window of company-specific eventsWhen ≥3 events accumulate in 7 days
Sector Analysis
Industry-wide trends using advanced taxonomyPeriodically

The Decoupled Trigger System

Why "decoupled"? Traditional systems wait for all data to be ready before analyzing. Our system lets different types of analysis run independently, so you get insights faster.

Foundational Triggers

Standard Procedure
New Report / Period Data

Deep, comprehensive analysis of the entire business and financial state.

Qualitative & Quantitative Synthesis

Incremental Triggers

Event-Driven
Event Accumulation (≥3) / Schedule

Rapid updates to layer in news, macro changes, and tactical events.

News, Macro & Weekly Intelligence

📄 Foundational Triggers

Triggered by new reports or data availability.

  • Create deep, comprehensive analysis
  • Take longer but produce complete picture
  • Form the backbone of assessments

⚡ Incremental Triggers

Triggered by event accumulation or schedules.

  • Keep analysis fresh between reports
  • Faster, more frequent updates
  • Capture breaking developments

Layer 2: The Investment Team (6 Agents)

Once Layer 1 has organized the data, Layer 2 agents work together like an investment committee to form a coherent view.

😊 Sentiment Analyst

Assesses overall tone across news, reports, and market signals.

"What's the mood?"

⚖️ Risk/Opportunity Analyst

Identifies top risks and opportunities with probability × impact scoring.

"What could go right/wrong?"

🔍 Data Quality Analyst

Evaluates freshness, completeness, and reliability of available data.

"Can we trust this analysis?"

🔄 Reconciliation Agent

Cross-checks qualitative claims against quantitative evidence. Exposes contradictions.

"Does the story match the numbers?"

📊 Charting Agent

Generates visual chart specifications to illustrate key thesis points.

"Show me the evidence"

🎯 Investment Orchestrator

Synthesizes all agent outputs into a final, coherent assessment. Makes the position call.

"The final decision"

How Assessments Work

The Investment Orchestrator assigns one of five positions based on the combined analysis.

PositionWhat It MeansTypical Conditions
Accumulate
Favorable risk/reward for adding to positionStrong fundamentals, reasonable valuation, positive catalysts
Hold
Current position appropriateMixed signals, fair valuation, stable outlook
Monitor
Watch closely, situation evolvingData gaps, pending catalysts, unclear direction
Reduce
Consider trimming positionDeteriorating fundamentals, elevated valuation risk
Avoid
Risk/reward unfavorableMaterial concerns, severe data gaps, regulatory issues

Transparency: What the AI Considers

Here are the key factors that influence our assessments:

Recent financial performance vs. expectations
Comparison to industry peers (standard taxonomy)
Market sentiment from news and events
Technical price patterns and momentum
Data quality and recency of information
Explicit contradictions between narrative and numbers
Macroeconomic factors affecting the sector
Management quality signals from reports

Limitations & Disclaimers

1.

AI is not a crystal ball. Our assessments are based on available data and may not anticipate unexpected events.

2.

Assessments are informational, not investment advice. We help you understand companies, not tell you what to do with your money.

3.

Always do your own research. Use our analysis as one input among many in your decision-making process.

4.

Past performance doesn't guarantee future results. Historical patterns may not repeat.