INTELLIGENCE ENGINE // 55 AGENTS // 7 TIERS

55 Agents. 7 Tiers.
One Verdict.

Every company passes through a forensic gauntlet. From financial statement analysis to corporate network mapping, our agents find what humans miss.

PHILOSOPHY

Forensic skepticism by default.

Agents ask, humans decide. Every signal is a question, not a conclusion. We present evidence -- you make the call.

The forensic override rule means that any single critical finding from a governance or forensic agent can override positive signals from all other tiers. A company with excellent financials but a promoter siphoning cash through related parties gets flagged.

We iterate with different minds. The same data is analyzed by agents with different perspectives -- a financial analyst, a forensic accountant, a regulatory expert, and a pattern matcher all examine the same company independently.

// forensic override rule
function verdict(company) {
  const tiers = runAllTiers(company);

  // Forensic override: any CRITICAL
  // from T2/T4/T6 overrides everything
  if (hasCritical(tiers.governance) ||
      hasCritical(tiers.regulatory) ||
      hasCritical(tiers.crossRef)) {
    return {
      verdict: "AVOID",
      reason: "forensic_override",
      confidence: 0.95
    };
  }

  return synthesize(tiers);
}

AGENT ARCHITECTURE

7 Tiers of Scrutiny

55 agents organized by domain. Each tier feeds the next.

T1

Core Financial

Revenue Quality Cash Flow Forensics Earnings Sustainability Balance Sheet Stress Working Capital Trap Capex Audit Debt Spiral Tax Anomaly
T2

Governance & Promoter

Promoter Behavior Board Independence Pledge Monitor Related Party Transactions Director Network KMP Turnover Auditor Red Flag Compliance Gap
T3

Market & Valuation

Valuation Disconnect Insider Trading Patterns Volume Anomaly Short Interest Signal Institutional Flow Bulk Deal Decoder Price-Earnings Forensics
T4

Regulatory & Legal

SEBI Watch NCLT Tracker MCA Filing Anomaly Penalty Monitor Debarment Check CIRP Status Consent Order Decoder
T5

Supply Chain & Operations

Customer Concentration Supplier Risk Inventory Buildup Geographic Exposure Capacity Utilization Contract Quality
T6

Cross-Reference & Pattern

Shell Company Detector Circular Trading Layered Entity Structure Common Director Network Address Clustering Capital Siphon Pattern
T7

Synthesis & Verdict

Synthesis Engine Historical Pattern Match Confidence Scorer Verdict Generator

Specialist Agents

XBRL Parser

Financial statement extraction

NLP Classifier

Disclosure language analysis

Network Mapper

Corporate relationship graphs

Timeline Reconstructor

Event sequencing

SIGNAL FRAMEWORK

High Frequency + Low Frequency

Different signals require different cadences. We run both.

HF

High Frequency

Signals that change daily or weekly. Insider trades, bulk deals, price/volume anomalies, SEBI orders.

  • Insider trade alerts
  • Bulk deal patterns
  • Price anomalies
  • New SEBI orders
LF

Low Frequency

Signals that emerge quarterly or annually. Earnings quality shifts, governance deterioration, balance sheet stress.

  • Earnings manipulation
  • Governance decay
  • Debt spiral
  • Promoter pledge creep

INTERACTION LOGIC

HF signals trigger immediate re-evaluation of LF conclusions. Example: A sudden bulk deal (HF) on a company already flagged for earnings manipulation (LF) escalates to CRITICAL instantly. The synthesis engine weighs both frequency bands to prevent alert fatigue while catching genuine escalation patterns.

LIVE DEMONSTRATION

Agent Feed Terminal

A real-time view of how agents analyze a company.

LIVE AGENT FEED MARUSHIKA TECH INDUSTRIES LTD
PROMOTER → MARUSHIKA TECH → CRITICAL

Longthrive Capital (Cayman) acquired 14.2% via offshore derivative. Promoter pledge rose from 22% to 67% in 3 quarters. Zero disclosure in AR.

FORENSIC → MARUSHIKA TECH → CRITICAL

DGGI investigation initiated. Cash flow from operations negative for 6 consecutive quarters despite 340% revenue growth. Receivable days: 287.

RPT_MONITOR → MARUSHIKA TECH → CRITICAL

Rs 17.5Cr parked in "Advance to Subsidiaries" — recipient entity incorporated 3 months prior. Common director: Vijay Mallesh (DIN: 09876543).

SYNTHESIS → MARUSHIKA TECH → AVOID

38 red flags across 7 tiers. Confidence: 0.94. Governance score: 12/100. Verdict: AVOID. Similar pattern match: Satyam (2009), DHFL (2019).

WHY ZENOLATTICE

The Competitive Edge

Agent Scale

55 specialized agents vs. single-model analysis. Each agent is an expert in its domain -- a promoter behavior agent doesn't care about valuation, it only tracks pledge patterns, stake changes, and related-party flows.

Forensic + Investment

Most platforms offer either forensic analysis or investment screening. We do both. The forensic override rule means red flags in governance can never be masked by attractive financials.

India Regulatory Depth

31 of our 55 agents are India-specific. They understand SEBI adjudication orders, MCA filing patterns, NCLT proceedings, and the unique dynamics of Indian promoter-driven companies.

RPT + Promoter AI

Related-party transactions and promoter behavior are the two biggest fraud vectors in Indian markets. We have dedicated agent clusters for each, cross-referencing across MCA, SEBI, and BSE data.

CASE STUDY

Marushika Tech Industries

VERDICT: AVOID 38 red flags | Confidence: 0.94

Offshore derivative acquisition by Longthrive Capital (Cayman) -- 14.2% stake with zero disclosure.

Promoter pledge escalation: 22% to 67% in three quarters.

DGGI investigation initiated. Cash flow from operations negative for 6 consecutive quarters despite 340% revenue growth.

Rs 17.5Cr parked in subsidiary incorporated 3 months prior. Common director identified.

Governance score: 12/100. Pattern match: similar trajectory to known fraud cases.

Intel access launching Q3 2026.

Request a demo of the intelligence engine.

Launch Q2 2026